This introductory chapter presents a review of the type 1 diabetes (T1D) literature including a historical perspective of diabetes, what is known about disease pathogenesis, and current efforts to identify genes for which allelic variation affects the risk for developing T1D. In parallel to ongoing work in humans, animal models have provided valuable insights into T1D pathogenesis and genetic susceptibility. The nonobese diabetic (NOD) mouse is currently the most extensively characterized model for T1D. This review also details investigations of the autoimmune and genetic basis for this disease in NOD mice, and the strategies used to identify genes underlying T1D susceptibility in this animal model.
Type 1 diabetes (T1D), also known as juvenile diabetes and insulin dependent diabetes mellitus (IDDM), is a disease resulting from the specific destruction of the insulin-producing pancreatic beta cells, predominantly mediated by lymphocytes. This leads to the deficiency of insulin in the body, in turn affecting normal blood glucose regulation. If untreated, this deficiency can lead to potentially fatal metabolic disorders including ketoacidosis. Hence, T1D is classified by its insulin dependency and autoimmune basis (1). Even with insulin therapy, serious complications (such as cardiovascular diseases, kidney failure and blindness) can arise, leading to mortality rate 20 times higher than the general population for those diagnosed during childhood (2). In 2003, it was estimated that T1D affects approximately 1,000,000 children under the age of 15years in the United States (3). Other estimates also reflect an increase (~3% per year) in the global incidence of this disease (4). However, T1D incidences vary from country to country (0.1 - 36.8 cases â‰¤14years/100,000/year), suggesting the involvement of environmental factors in disease pathogenesis (5-7). With the increasing occurrence of T1D, concerted efforts have been focused on determining the factors contributing to the autoimmune destruction of the pancreatic beta cell.
1.2 A historical perspective of the 'sugar disease'
Get your grade
or your money back
using our Essay Writing Service!
Records for the clinical features currently associated with diabetes were first recorded in ancient civilization. Nearly 3,500 years ago, an Egyptian physician, Hesy-Ra, noted the passing of excessive urine as an ailment (8). The name 'diabetes' (derived from the Greek word for "siphon") was first coined by Aretaeus approximately 1,500 years later, to describe the abnormal passage of large volumes of urine (9). During the early 1800s, the term 'mellitus', Latin for "honey", was added to diabetes when abnormally high concentrations of sugars were found in the urine of disease-afflicted individuals (8). The importance of the pancreas in this sugar disease was only demonstrated some 90 years later by Minkowski and von Merhing, when diabetes developed in pancreatectomized dogs (10, 11). Further pathological examination then associated diabetes with the 'degeneration' of the pancreatic islets of Langerhans (12). These observations led to the deduction that the pancreas produced an 'internal secretion' implicated in carbohydrate metabolism (10, 11).
Many investigators set out to isolate the internal secretion of the pancreatic islets. The name 'insuline' (derived from the Latin word for "island") was subsequently suggested for this internal secretion, because it describes the tissue from which it was derived (13, 14). After numerous unsuccessful attempts by others, Banting and his assistant Best in 1921 isolated a pancreatic extract that was able to relieve glycosuria in pancreatectomized dogs (13, 14). In collaboration with Collip and Macleod, they managed to purify the internal secretion of the pancreas and renamed it 'insulin'. The team then demonstrated that diabetic patients, upon insulin delivery, showed improvements in glucose regulation. Subsequently, insulin was biochemically localized to the beta cells present in the pancreatic islets (15). The landmark discovery of insulin and its clinical applications led Banting and Macleod to be presented the Nobel Prize in Physiology or Medicine in 1923 (14). To date, individuals afflicted with diabetes still rely on the regular administration of insulin to treat this potentially fatal disease (16).
1.3 The autoimmune basis of diabetes
Initially, the mechanism of pancreatic islet cell destruction, which now defines T1D and results in the loss of insulin, was unclear. One of the earliest hints of the autoimmune aetiology of T1D was through observations by Schmidt in 1902. He described the presence of mononuclear cell infiltration in the pancreata of individuals with diabetes (17). Further histological examination by others confirmed this observation by demonstrating the presence of lymphocytic infiltrates in the pancreatic islets and the loss of beta cells in patients recently diagnosed with the disease (18, 19). These reports led to investigations into this apparent immune hypersensitivity towards the islets of Langerhans during diabetes pathogenesis.
Always on Time
Marked to Standard
The pancreas-specific infiltration, as well as the presence of anti-islet cell antibodies (ICAs), provided direct evidence for the autoimmune basis of diabetes. Initially, Nerup et al. observed lymphocytes specific for beta cell antigens, infiltrating the pancreatic islets of patients with diabetes (20). In a subsequent report, T lymphocytes, in particular CD8+ T lymphocytes were demonstrated to be the predominant component of these lymphocytic infiltrates (21). The autoimmune hypothesis for diabetes pathogenesis was further supported when two groups found ICAs in diabetic individuals concomitant with other autoimmune disorders such as vitiligo and thyrotoxicosis (22, 23). Several beta-cell proteins, including (pro)insulin (24-28), insulinoma-associated protein 2 (IA-2) (29-33), glutamic acid decarboxylase (GAD) (34, 35), islet-specific glucose-6-phosphatase catalytic subunit related protein (IGRP) (36) and zinc transporter 8 (ZnT8) (37) have now been shown to be recognized by T and B lymphocytes isolated from diabetic patients (reviewed in (38, 39)). Of these autoantigens, proinsulin appears to play a critical role in T1D onset (40). The detection of high titers of antibodies specific for these beta-cell autoantigens have since been used to identify individuals at risk for subsequent development of T1D (41-43).
A number of therapies targeting the autoimmune response have been, or are currently, being evaluated as potential treatments for T1D (reviewed in (44, 45)). These clinical trials largely involve either the use of antibodies to deplete specific lymphocyte subsets or beta-cell autoantigens to modulate the immune response. One example is the use of modified anti-CD3 monoclonal antibodies (mAb) administered to recent-onset T1D patients and/or individuals with a high risk of developing T1D (46-48). It was anticipated that the administration of these modified antibodies would dampen effector T- cell activation and prevent/delay disease onset. However, it was noted that a number of T1D patients involved in the anti-CD3 mAb trial developed transient Epstein-Barr virus reactivation (49).
Trials using antigen-specific immune therapies have also been conducted using beta-cell autoantigens, such as insulin and GAD, in patients highly susceptible to T1D development (50-53). The overall aim of administering these autoantigens is to induce anergy in autoreactive T cells and antigen-specific immunosuppressive responses from regulatory T cells. Nevertheless, current immune therapies have had limited success in preventing T1D development, and some have even provided unexpected adverse effects in at-risk patients (49, 51-53). Ongoing efforts have been focused on fine-tuning these therapies, as well as developing new combinatorial strategies to prevent T1D progression (reviewed in (44)). In the meantime, numerous questions regarding the etiological factors that trigger autoimmune initiation in T1D are still left unanswered - understanding these factors may reveal new targets for designing/optimizing treatments for this disease.
1.4 Genetics of T1D
Genetic studies have identified inherited factors contributing to the increased familial risk of T1D. For example, an individual with a T1D affected sibling has ~6% risk of developing T1D compared to someone in the general population who has a 0.4% risk (54). However, the concordance rate between monozygotic twins has been reported as less than 60% (55, 56), suggesting that susceptibility genes act in concert with environmental factors to initiate disease pathogenesis (reviewed in (57, 58)).
1.4.1 Approaches for identifying disease susceptibility genes
Genetic predisposition to T1D is largely determined by complex interactions between multiple chromosomal loci, conventionally termed IDDM loci to denote insulin-dependent diabetes mellitus susceptibility (reviewed in (59)). With the use of polymorphic molecular markers, such as single nucleotide polymorphisms (SNP), insertions/deletions and nucleotide repeats, two main approaches have been used to identify these T1D susceptibility loci: linkage and association studies.
Linkage studies are typically used to determine the sharing of alleles for genetic markers within families having more than one affected child. The affected sib-pair method is a common statistical linkage analysis used to identify chromosomal regions contributing to disease susceptibility ((60) and reviewed in (61-63)). This method involves genotyping genetic markers at specific positions or across the genome for sibling pairs with T1D, and evaluating the proportion of allele sharing for each marker. Linkage of a marker to T1D is indicated by significant excess allele sharing (identical by descent from the parents and greater than would be expected under independent assortment) within a cohort of affected sibling pairs from different families.
Association studies, on the other hand, were initially used to test genetic markers between groups of individuals representing cases (i.e. individuals with T1D) and unrelated controls (i.e. individuals without T1D). Traditionally, these studies were performed by determining the frequency of particular alleles of candidate genes in a population of affected individuals compared to unaffected controls. Recently, instead of testing genes individually, large marker sets covering almost the entire genome are simultaneously genotyped without prior selection of candidate genes - such studies are termed as genome-wide association studies (GWAS, reviewed in (62)). The frequencies of these marker alleles are then compared between patients with T1D and either healthy, unrelated individuals (case-control) or amongst related individuals (family-based). Using DNA repositories from previous linkage analyses, the transmission disequilibrium test (TDT), a widely-used family-based method, can be used to assess the transmission frequencies of inherited alleles within families with affected individuals (reviewed in (64, 65)). TDT assumes that a parent heterozygous for a marker linked to disease will more frequently transmit the associated allele than an un-associated allele to an affected child. Association analyses have been used to establish additional evidence for loci identified by linkage analysis and the possibility of identifying causative alleles for T1D susceptibility candidate genes.
1.4.2 Disease susceptibility loci, genes identified and limitations of genetic studies
This Essay is
a Student's Work
This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.Examples of our work
The first susceptibility locus for T1D mapped within the human leukocyte antigen (HLA) region (termed IDDM1, (60, 66-70)). This region is the major histocompatibility complex (MHC) in humans, and is found on chromosome (Chr) 6p21. Specifically, T1D is associated with certain HLA class II alleles (71, 72). This suggests a critical role of HLA class II in influencing peptide presentation for the activation of autoreactive T lymphocytes (73, 74). The evaluation of risk factors for relatives of patients with T1D estimated that HLA-identical siblings have up to 17% risk, while an individual with an affected, genetically identical twin have up to 50% risk of developing the disease (55, 56). This increase in risk for genetically identical twins (17% to 50%) suggests that non-HLA loci, along with environmental factors, must also contribute towards T1D susceptibility.
Soon after the discovery of IDDM1, the first non-HLA T1D susceptibility locus was identified. This locus maps to a variable number tandem repeat (VNTR) 5' of the insulin gene (INS), and is located on Chr11p15 (IDDM2, (75-77)). Allelic variation in the length of the VNTR was proposed to alter the chromatin structure and impede the accessibility of transcription factors (78-80). These polymorphisms were correlated with INS expression in the thymus and pancreas, whereas short VNTR alleles (26 - 63 repeats) were found to provide the highest disease risk (76, 81, 82). Differences in thymic INS transcription have been suggested to affect T cell tolerance induction, which in turn can influence susceptibility to autoimmunity and beta cell destruction (83).
IDDM1 and IDDM2 were identified using candidate gene analysis due to the likelihood of HLA genes and insulin being involved in T1D pathogenesis. To date, genome-wide linkage analyses have mapped 19 other non-HLA loci shown to contribute to T1D susceptibility, suggesting that this disease is polygenic (reviewed in (57)). However, some loci did not provide significant evidence for linkage to T1D susceptibility in every population tested (e.g. IDDM3, IDDM5 and IDDM7 (84-88)). Even when confirmed in multiple populations, chromosomal linkage generally resulted in relatively poor mapping resolution (>100 genes in a linked region, e.g. IDDM4) or in a region where there are no candidate genes (e.g. IDDM8, IDDM9 and IDDM10). Hence, investigators looked to association-based approaches to provide higher mapping resolution for identifying disease susceptibility genes within these linked regions.
Candidate genes for several T1D susceptibility loci have been confirmed or eliminated based on association studies. IDDM2/INS, which initially demonstrated relatively weak linkage in previous analyses, was confirmed when this locus showed strong association in family-based association studies (77, 89). For other regions, candidate genes were chosen based on their immunological role(s). CTLA4, critical for T cell signal regulation, maps to Chr2q33 (IDDM12) and SUMO4, important for regulating nuclear factor-ÎºB (NF-ÎºB) signalling, maps to Chr6q25 (IDDM5) (90-94). Alleles for both genes were found to be inherited at higher frequencies in affected individuals. Association-based studies also strongly supported the possible involvement of two other candidate genes, PTPN22 and IL2RA/CD25 (95-97). These two genes encode a tyrosine phosphatase which suppresses T cell activation, and a cell surface molecule involved in regulating T cell proliferation respectively (reviewed in (98, 99)). In contrast, previous candidates such as GAD65, FOXP3, LCK, FADD, LRP5 and CBLB were ruled out as significant preferential transmission of alleles to affected individuals was lacking in various studies (100-105). Moreover, the contribution of a candidate gene can be unclear with allelic association found in some populations, but not others. For example, IL12B (IDDM18)Â demonstrated allelic association with T1D in collections from Australia, Japan, Spain, UK and USA (106-109), but not in other collections from France, Italy, Northern Ireland, Norway, Scandinavia, and USA (110-113). Therefore, genetic evidence for a particular candidate gene may be strong due to association within several populations tested, or restricted to a specific population. Then again, limited association can be influenced by a small sample size and may not be validated in larger studies of the same population. In either case, these initial genetic association studies for T1D typically focused on particular chromosomal regions or candidate genes of interest.
To simultaneously test marker alleles across the entire genome, multiple groups and consortiums have more recently performed GWAS using high-density SNP genotyping arrays to identify genes/regions contributing to T1D risk (114-119). These studies provided further validation for previously identified candidate genes - INS, CTLA4, PTPN22, IL2RA, and revealed >30 novel loci contributing to T1D risk. The majority of associated SNPs appear to occur near genes that have reported functions in immune responses (59). Notably, several of these SNPs are located within or in close proximity to genes implicated in T cell function (e.g. CD69, PTPN22 and UBASH3A (120-122)). These genes are promising candidates because T cells are a dominant cell type of the pancreatic lymphocytic infiltrate. In addition, a number of other candidate genes were found to be highly expressed in the pancreatic beta cells (e.g. COBL, CTSH and PRKD2 (123)). Given that the pancreatic beta cells are specifically destroyed during T1D progression, these genes may be essential in insulin production and/or maintaining beta cell viability. Hence, GWAS have revealed useful insight into the genetic variants and pathways that may be critical in T1D pathogenesis.
Despite the efficiency of GWAS in confirming previous as well as revealing new T1D susceptibility loci, these studies were often unable to identify a single gene within a region associated with T1D. Association was typically detected across several markers within a chromosomal region harbouring many genes (i.e. multiple SNPs provided statistically significant association scores). For example, allelic association for markers within Chr12q24 were associated with disease susceptibility and spanned a >1.5 Mb region (124). This region contains at least 15 known genes which make it challenging to identify a specific gene, let alone pinpoint the causative variant contributing to T1D susceptibility. Even when confirmatory support was provided for a candidate gene, allelic variation within these genes appear to provide small effects upon disease risk - compared to allelic variation for HLA, which has an odds ratio (OR) of ~6.8, allelic variation for previously identified candidate genes provided ORs of <2.5 (reviewed in (125)). Furthermore, it appears that the current collection of associated genomic regions do not adequately explain the overall heritability of disease predisposition (reviewed in (126)). Therefore, while GWAS have identified numerous loci contributing to disease risk, this strategy has elicited mixed feelings regarding the success of this strategy for pinpointing genes contributing to T1D susceptibility.
In general, current human genetic strategies can be hindered by different factors when trying to map genes for which allelic variation contributes to T1D predisposition. One factor is genetic heterogeneity, where different alleles or combinations of alleles for different genes can contribute to greater or lesser susceptibility in different populations or due to the influence of environmental factors (reviewed in (63)). For example, certain common variants on their own may provide subtle effects, but in the presence of other variants or environmental factors may provide a more profound effect upon T1D risk. Statistical power for these strategies represents another limiting factor (reviewed in (59)). Due to the nature of linkage analyses, only loci with large effects upon disease risk will be detected unless extraordinarily large cohorts of affected sib-pairs are collected and genotyped. Current GWAS, on the other hand, have focused on common variants and thus have only identified association of disease with common variants, which often contribute small effects to complex genetic diseases. Instead, it has been proposed that rare alleles that could not be detected by recent GWAS, may provide larger effects and account for the missing genetic variance (127). At present, strong genetic evidence has been provided for T1D susceptibility genes, but it is uncertain whether the actual causal variants for the majority of these genes will be identified, let alone their effect upon T1D pathogenesis determined.
1.5 The NOD mouse: A model for T1D
As an alternative approach which avoids the complexity of human studies is identification of T1D susceptibility genes within mouse strains. Besides fewer constraints on tissue availability for pathological analysis, mouse strains enable selective breeding in a controlled environment to pinpoint the allelic variants contributing to T1D susceptibility. To date, the most extensively characterized model for T1D is the NOD mouse strain. This strain was discovered by chance during a selective inbreeding experiment of "Swiss" mice to isolate a strain that develops cataracts (reviewed in (128)). At the F6 generation, mice which exhibited elevated fasting blood glucose levels were selectively inbred. The progeny of these mice progressively led to the derivation of the NOD mouse strain which spontaneously develops hyperglycaemia and glycosuria. Similar to human T1D, disease predisposition in this mouse model has a polygenic nature (129). Hence, the NOD mouse has been used to identify susceptibility genes and study the complex genetic interactions contributing to T1D onset.
Disease pathogenesis in the NOD mouse has an autoimmune basis similar to that observed in human T1D (reviewed in (130)). Two major phases have been identified for the progression of pancreatic beta cell destruction. Insulitis, the first phase, occurs when a stochastic event triggers leukocytes to recognize self-antigens and infiltrate the pancreatic islets. Analogous to humans, these self-antigens include insulin, IGRP and GAD, with responses to insulin being pivotal to disease development (131-136). The leukocytic infiltrate then initiates an inflammatory cascade which progressively mediates the specific destruction of beta cells. Mild insulitis can be detected in NOD females by 4-5 weeks of age and complete insulitis is present in ~100% of mice by 30 weeks of age. Diabetes, the second phase, ensues when the leukocytic infiltrate progressively accumulates and destroys enough beta cells causing insufficient insulin production. Consequently, metabolic disorders such as hyperglycaemia develop following disruptions in glucose homeostasis (137).
Although NOD mice exhibit all the distinctive symptoms of T1D, slight differences do exist between this mouse model and disease pathogenesis in humans. Unlike humans, NOD mice present a strong sex bias in T1D development. By 30 weeks of age, NOD females are at ~70%-90% risk while NOD males are at ~20%-50% risk (138). In addition, this strain does not display ketoacidosis, a characteristic complication of untreated human T1D patients, due to the ability of mice to metabolize ketone bodies in the liver (139). Despite these minor differences, NOD mice remain one of the best tools for studying susceptibility genes and initiating factors contributing to autoimmune T1D pathogenesis.
1.6 Autoimmune basis of T1D in NOD mice
The hallmark of T1D in NOD mice is the specific destruction of insulin-producing beta cells in the pancreatic islets mediated by lymphocytes. While the initiating factors that trigger this autoimmune response is currently not well-understood, a number of different mechanisms have been proposed. A predominant model suggests that when the beta cell mass undergoes remodelling at approximately 2 - 3 weeks of age, beta cell antigens, such as insulin, are released during the ensuing wave of apoptosis or necrosis (140-143). These antigens become available to antigen presenting cells (APC), which are then able to present and stimulate diabetogenic lymphocytes. Another model proposes the involvement of microbial and viral agents which directly activate beta cell-specific lymphocytes or indirectly via molecular mimicry (reviewed in (144-147)). In a more recent study, TRPV1+ sensory neurons have been implicated as the third model for the initiation of beta cell destruction. These neurons innervate the pancreas and induce lymphocytic infiltration into the beta islets (148). In either of the three models, activated autoreactive lymphocytes consequently migrate to the pancreas and destroy the insulin-producing beta cells.
1.6.1 Cellular mediators of antigen presentation and beta cell destruction
In order to become activated, naÃ¯ve autoreactive lymphocytes will first have to recognize self-peptides bound to MHC complexes on the surface of APCs (reviewed in (149)). Dendritic cells (DCs) and macrophages have been demonstrated histologically to be the first APCs to infiltrate the damaged islets (150). The critical role of these immune cells in initiating diabetes progression was highlighted when NOD mice with inactivated macrophages were unable to transfer disease (151). NOD DCs (discussed in Section 1.6.3) and macrophages have also been shown to be the first key producers of TNF-Î±, a proinflammatory cytokine, supporting the involvement of these APCs early in the T1D development (152). NOD macrophages have additionally been reported to produce large amounts of other proinflammatory cytokines including, IL-12 and IL-1Î², following exposure to apoptotic or necrotic cells (153, 154). This aberrant cytokine secretion may enhance the recruitment of precursor DCs and monocytes to the beta cell lesion, further promoting autoimmunity (155-157).
Upon antigen engulfment, APCs migrate to the local lymph nodes (i.e. pancreatic lymph nodes, PLN) where they present beta cell antigens to lymphocytes (reviewed in (158)). Antigen presentation in the PLN is a critical step in T1D pathogenesis because this is where diabetogenic lymphocytes encounter self peptide/MHC complexes (159, 160). The absence of this tissue was able to almost impede diabetes development in NOD mice (160, 161). In the unfortunate event when an autoreactive lymphocyte recognizes a peptide/MHC complex on the surface of an APC, this lymphocyte will become activated and migrate to the pancreatic islets (reviewed in (162)). Although the absolute requirement of diabetogenic lymphocytes to engage self antigens in the PLN in order to become activated has been recently contested, activated autoreactive lymphocytes ultimately undergo clonal expansion and are present in the pancreatic islets to progressively destroy the insulin-producing beta cells (163, 164).
Two main cellular mechanisms of beta cell destruction, predominantly mediated by T lymphocytes, have been proposed (158). First, autoreactive CD8+ (cytotoxic) T lymphocytes cause the initial pancreatic beta cell destruction through direct cell contact and the release of perforin and granzymes following Fas/FasL interaction (165). The second mechanism is via the release of pro-inflammatory cytokines including IFN-Î³ and TNFÎ± by autoreactive CD4+ T lymphocytes and macrophages present in the pancreatic beta cell milieu (153). In either case, T lymphocytes have been recognized to be main effectors of beta cell destruction.
Besides T lymphocytes, B lymphocytes have also been shown to play critical roles in T1D pathogenesis (reviewed in (166, 167)). A deficiency of B lymphocytes in NOD mice prevented T1D development, and this protective effect was reversed when these mice were repopulated with NOD B lymphocytes (168-171). B lymphocytes are able to produce autoantibodies specific for beta cell autoantigens such as insulin, GAD and IA-2 (172). However, the autoantibody production role of B lymphocytes in T1D development has been suggested to be a consequence of beta cell destruction because the transfer of these autoantibodies into B lymphocyte-deficient NOD mice did not affect T1D resistance (170). On the other hand, B lymphocytes have also been shown to be efficient APCs in the activation of diabetogenic CD4+ T lymphocytes during the initiation of T1D onset (173, 174). Specific B lymphocyte clones were reported to preferentially recognize beta cell proteins via surface antigen-specific immunoglobulin molecules, and trigger diabetogenic CD4+ T lymphocyte activation. Increasing the levels of particular beta cell antigen-specific B lymphocyte clones in NOD mice provided T1D exacerbation (175). Finally, recent work also suggested that B cells provide pro-survival signals to cytotoxic T lymphocytes present within the inflamed islets (176). Therefore, T and B lymphocytes recruited to the beta cell lesion are both essential in escalating the immune response leading to pancreatic beta cell destruction.
1.6.2 Cellular mediators of an immune response, what could go wrong?
Given that T lymphocytes are the key contributors towards T1D pathogenesis, numerous studies have focused on the development of these cells in the thymus (reviewed in (162)). During negative selection, thymocytes with strong reactivity towards self-antigens are eliminated, preventing them from being released into the periphery (177). The requirement for thymic beta-cell autoantigen expression for T1D protection in NOD mice is currently still under contention. Decreased thymic insulin expression in these mice resulted in diabetes exacerbation, while a deficiency or overexpression of GAD65 did not affect diabetes susceptibility (178-180). Defects in the elimination process involving Fas-dependent and -independent apoptotic pathways, have also been observed in NOD mice (181, 182). In another study, the resistance of developing thymocytes to thymic deletion in NOD mice was attributed to a failure to induce expression of Bim, a proapoptotic gene (183). These defects in combination may allow more self-reactive NOD thymocytes to circumvent central tolerance mechanisms and escape into the periphery.
Peripheral T lymphocyte numbers are stringently regulated at a steady state in diabetes-resistant strains (184). In order to maintain T-lymphocyte numbers, it has been proposed that naÃ¯ve T lymphocytes (including autoreactive T lymphocytes) undergo an expansion referred to as homeostatic proliferation (reviewed in (185, 186). Autoreactive T lymphocytes may be favoured during this expansion process which may promote T1D onset in NOD mice (187). Such an event was highlighted in a study performed by King et al. who reported that naÃ¯ve T lymphocytes specific for beta-cell autoantigens revealed a greater propensity to proliferate in lymphopenic NOD mice compared to mice with a 'filled' lymphoid compartment (187). The extent of this autoreactive T lymphocyte proliferation was found to be proportionate to the severity of insulitis in these mice. Then again, the enhanced ability of NOD T lymphocytes to proliferate in a lymphopenic environment was not observed in another study (188). While the lymphopenic state of NOD mice has been controversial, increased levels of autoreactive T lymphocytes in the peripheral immune repertoire may predispose these mice to T1D progression.
Under normal circumstances, regulatory cell subsets in the periphery can induce tolerance in autoreactive lymphocytes to prevent the initiation of autoimmune disease (189). NOD mice exhibit deficiencies in four well-characterized immunoregulatory cell populations - CD4+CD25+ T cells (190, 191), natural killer T (NKT) cells (192, 193), natural killer (NK) cells (192), DCs ((194-196), discussed in Section 1.6.3) and more recently CD4(-) CD8(-) T cells (197). Although the levels of CD4+CD25+ T cells appear to vary between colonies, increasing the levels of CD4+CD25+ T cells and NKT cells in NOD mice either by stimulation or adoptive transfer was able to delay or prevent T1D onset (188, 191, 198-203). Besides the levels of these regulatory cell subsets, abnormalities in the characteristics and function of these cells have also been reported. For example, CD4+CD25+ T cells in NOD mice have been shown to express reduced levels of cell surface adhesion molecules (204). NOD NKT cells, on the other hand, were unable to secrete inflammatory cytokines following T cell receptor engagement (193, 205). Further, an NK cell export defect was also suggested when increased numbers of NK cells were found in the bone marrow of NOD mice (192). Abnormalities in these NOD peripheral immunoregulatory cell subsets, may consequently allow diabetogenic lymphocytes to mediate the specific destruction of the insulin-producing pancreatic beta cells.
1.6.3 DCs: Roles in immunity and tolerance
DCs have been implicated in both initiating an immune response, as well as inducing tolerance in autoreactive lymphocytes (reviewed in (206, 207)). At steady-state, these cells exist at an 'immature' state and are highly efficient in maintaining immune surveillance by constantly engulfing and presenting self (released by apoptotic cells) and foreign proteins (released by pathogens when present) to the T lymphocyte repertoire. However, T lymphocytes typically do not become activated upon encounter with an immature DC (208-210). Upon receiving activation stimuli, DCs 'mature' and upregulate co-stimulatory molecules (CD80, CD86 and CD40) required for T lymphocyte activation. These features of DC development hence enable these cells to regulate immune responses towards self, as well as invading pathogens.
Within the heterogeneous population of DCs, the two main DC subsets implicated in T1D pathogenesis are conventional dendritic cells and plasmacytoid dendritic cells (194, 196, 211, 212). In the absence of inflammation, conventional dendritic cells (cDCs) are found resident within primary and secondary lymphoid tissues (reviewed in (213)). Two cDC subsets that are common to all lymphoid tissues have also been identified: CD8+ and CD8- cDCs. Notably, CD8+ cDCs have a unique ability to cross-present exogenous antigens and induce T lymphocyte activation (214, 215). This ability enables this DC subset to be efficient in maintaining peripheral T lymphocyte tolerance (216). NOD mice have been shown to be deficient in these cells, but correction of this abnormality either by in-vivo stimulation or by transferring CD8+ cDCs prevented or delayed diabetes onset respectively (194, 217, 218). CD8- cDCs have only been reported to produce high levels of pro-inflammatory chemokines (219). Since multiple other immune cells also express CD8 on their surface, another cell surface molecule, signal regulatory protein-alpha (SIRPÎ±), has been used in combination with CD8 to clearly define CD8+ and CD8- cDC subsets (220). Plasmacytoid DCs, on the other hand, continually circulate between lymphoid tissues via the blood (reviewed in (221, 222)). These cells produce vast amounts of type I interferon (IFNÎ±/Î²) following viral stimulation.
DCs require interaction with molecules found specifically on pathogens (pathogen associated molecular patterns, PAMPs) in order to mature. These PAMPs are recognized by pathogen recognition receptors such as Toll-like receptors (TLR). To date, at least eleven different TLRs which recognize particular PAMPs have been described (reviewed in (223)). For example, TLR4 and TLR5 respectively bind lipopolysaccharides (LPS) and flagelin, molecules present within Gram-negative bacteria, while TLR3, TLR7 and TLR9 recognize viral and bacterial nucleic acids. DCs (and macrophages) generally express most TLRs, but specific DC subsets selectively express certain TLRs. CD8+ cDCs express high levels of TLR3, while pDCs exclusively express TLR7 and TLR9. Since viral and bacterial components have been implicated as potential triggers of T1D pathogenesis via molecular mimicry, a number of groups have investigated the effect of compounds which mimic the effects of these components upon disease onset (224-227). However, these studies revealed conflicting results. NOD mice treated with cytosine-phosphodiester-guanine oligodeoxynucleotides (CpG), a bacterial DNA mimetic prevented T1D development in one study (228), but not in a more recent study (229). These controversial observations were similarly observed for NOD mice treated with polyinosinic:polycytidylic acid (PolyI:C), a viral double-stranded (ds) RNA mimetic (225-227). Thus, it is currently still unclear which TLRs are involved, and how the stimulation of these receptors provide a balance between immunity and autoimmunity.
Although certain DC subsets can directly tolerize autoreactive T lymphocytes, it has been proposed that DCs may also indirectly regulate these lymphocytes via regulatory T cells (230). For instance, DCs have been shown to affect the levels of CD4+CD25+ regulatory T cells (203, 231). In addition, immature CD8+ cDCs have been demonstrated to produce high levels of transforming growth factor (TGF) Î², which induces the expression of a regulatory T cell-specific transcription factor, FoxP3, in regulatory T cell precursors (232, 233). In either case, NOD mice exhibit deficiencies in a number of DC subsets, and increasing the levels of these cells prevents T1D onset (194-196). DCs in these mice also present maturation defects such as decreased expression of co-stimulatory molecules and a reduced capacity to stimulate T cells (234, 235). Therefore, abnormalities in the level, development and function of various immune cell subsets may be crucial in predisposing NOD mice to T1D development. More importantly, deficiencies and abnormalities in regulatory cell subsets (NK cells, NKT cells, CD4+CD25+ T cells and DCs), comparable to that observed in the NOD mouse, have been found in patients or individuals at risk for T1D (236-239). These observations further support the NOD mouse as a good model for human T1D.
1.7 Genetics of T1D in NOD mice
Similar to T1D in humans, multiple loci are linked to T1D susceptibility in the NOD mouse genome (reviewed in (240)). These loci have been conventionally termed Idd to denote mouse insulin-dependent diabetes mellitus susceptibility locus (note: it was recently suggested that these loci be identified by the chromosome location and the diabetes-resistant strain which aided its discovery, but to limit confusion, the conventional Idd nomenclature will be used in this thesis (241, 242)). Identical to IDDM1 in humans, Idd1 was the first susceptibility locus identified in NOD mice using a candidate gene approach. This locus contains the MHC gene region and provides the largest genetic contribution for T1D susceptibility in NOD mice (243-246). In particular, the MHC H-2g7 haplotype, unique to NOD mice, includes the MHC class II I-Ag7 allele (reviewed in (247)). Analogous to the human diabetogenic DQB1*0302 allele (HLA-DQÎ²), the I-Ag7 allele results in conformational changes to the peptide binding groove (248-250). Besides the I-Ag7 allele, the H-2g7 haplotype also includes a deletion in the promoter region of the I-EÎ± gene, resulting in the lack of surface I-EÎ± expression (251). The introduction of a resistant I-A allele or a functional I-EÎ± gene into NOD mice was able to prevent disease onset, confirming the contribution of these alleles to T1D progression (243-246). However, T1D did not develop when the MHC gene region from NOD mice was introduced into a diabetes-resistant C57BL/6 (B6) genetic background, suggesting the involvement of allelic variation within other genes that contributes to T1D susceptibility (252).