NOD is a descriptor string denoting the nonobese diabetic inbred strain developed by inbreeding of ICR (Swiss) mice in Japan (9). The strain is best referred to as NOD, since not all NOD mice will develop diabetes, although most mice of this strain develop insulitis. The discovery of these mice emphasizes the role of serendipity in science. The NOD and certain of the NOD-related inbred strains were developed from Jcl:ICR progenitors at the Shionogi Research Laboratories in Aburahi, Japan, by Dr. S. Makino (8,9,10). Selection for dominant cataract with microphthalmia was the initial objective of the breeding program in Makino’s laboratory. This selective breeding program led to the development of the CTS (cataract Shionogi) strain, in which all mice develop cataracts, and of NCT, a cataract-free control strain. At the sixth generation of inbreeding (F6), Makino noticed that some mice that were free of cataracts exhibited high fasting blood glucose levels. By selective breeding of F6 sibs and their progeny that exhibited this phenotype, Makino hoped to develop a new inbred strain exhibiting spontaneous diabetes. At the same time, he recognized the need for a euglycemic control strain and simultaneously inbred F6 sibs and their progeny that exhibited normal fasting blood glucose levels. In 1974, at the 20th generation of inbreeding (F20), a female spontaneously developed overt type 1 diabetes associated with heavy leukocytic infiltrations within the pancreatic islets (insulitis). Paradoxically, this female was not found in the line being selected for fasting hyperglycemia. Instead, the female with spontaneous type 1 diabetes was found in the line being inbred as a diabetes-free euglycemic control line. The progeny of this diabetic female were the founders of the NOD strain. Since the line originally selected for high fasting glucose levels never progressed to overt diabetes at the time that type 1 diabetes appeared in the euglycemic control (the latter now the NOD) line, the former strain was designated NON, for nonobese normal, and more recently, redesignated as nonobese nondiabetic (8,11). It was serendipitous not only that Makino coselected a euglycemic “control” line but also that his screening of this line was sufficiently rigorous to permit the discovery of that the “exceptional” female and the selection of her progeny to produce the NOD strain. Another serendipitous happenstance was that the husbandry environment in Makino’s vivarium was sufficiently free of murine pathogenic agents such that the diabetic phenotype could develop (see the section “Critical Role of Environment”). It should be noted that while subsequent inbreeding of ICR mice in Japan produced other inbred strains, including the ILI (ICR-L line-Ishibe), NSY (Nagoya-Shibata-Yasuda), ALS (alloxan-induced diabetes susceptible), and ALR (alloxan-induced diabetes resistant), none of the latter have developed spontaneous, autoimmune type 1 diabetes (although some that will be described in sections to follow
have developed IGT, pre-type 2 diabetes, or overt type 2 diabetes) (12,13,14).

Distribution of NOD and its closely related strains was tightly restricted to Japan within the first 5 years of their development (15). However, NOD mice are now widely available from suppliers in Japan (NOD/Shi from Clea, Tokyo), the United States (NOD/LtJ from The Jackson Laboratory, Bar Harbor, ME; and NOD/MrkTac from Taconic, Germantown, NY), and Europe (NOD/HlBom, Bomholtgard, Ay, Denmark). Substrain differences have been reported, notably, the very low incidence of type 1 diabetes present in the NOD/Wehi substrain in Australia (16). To avoid substrain divergence resulting from accumulation of recessive mutations distinct from those being fixed in the source colony, new breeders from the source colony can be imported every 10 to 20 generations to maintain uniformity with the mice being studied by other investigators.

Investigators studying NOD mice must be aware of the critical role of a specific pathogen-free environment in allowing penetrance of the underlying genetic susceptibility (17). The investigator has to choose between two options: maintaining a NOD breeding colony on site or ordering the requisite number of mice necessary for specific studies. Whichever option is selected, the investigator must house the mice in a specific pathogen-free (SPF) vivarium maintained to high standards of cleanliness and good animal care practice to achieve the diabetes frequencies mentioned above [reviewed in reference (18)]. This is necessary because NOD mice are subject to protective immunomodulation when stimulated by a broad spectrum of extrinsic pathogens, including viruses, bacteria, or their products or when fed semipurified diets lacking diabetogenic catalysts (6).

In marked contrast to the male gender-biased susceptibility observed in rodent models of type 2 diabetes to be discussed later, one of the mouse-specific characteristics of type 1 diabetes pathogenesis in NOD mice is that it is female gender-biased. This bias is presumably attributable to the development of larger lymphoid organs in females than in males and the stronger response of females to immunization. In high-incidence colonies of NOD mice, the incidence in females typically reaches 80% to 90% by 24 weeks of age, whereas the incidence in males is lower and much more variable, usually reaching more than 40% by 30 weeks of age in SPF colonies. In the author’s (EHL) colony of NOD/Lt males at The Jackson Laboratory, the frequency of type 1 diabetes in males varies from year to year, with a recorded low of 40% by 30 weeks of age and a high of 70% by 24 weeks of age. One of the unusual characteristics of the development of type 1 diabetes in NOD mice is that insulin therapy is not obligatory to sustain life over the first 2 to 4 weeks following the first diagnosis of hyperglycemia or glycosuria. This, in part, represents the ability of diabetic NOD mice to resist development of severe ketoacidosis (19). Diabetic NOD/Lt mice in the author’s colony must exhibit chronically high blood glucose levels (>500 mg/dL) before severe ketonemia can be demonstrated. For example, a diabetic NOD/Lt female with a glucose level of 337 mg/dL had a β-hydroxybutyrate level of only 0.5 mg/dL, a nondiabetic background level. A dipstick test of this mouse’s urine was also negative for ketones (acetoacetic acid). Two chronically diabetic NOD/Lt females (glucose levels of 913 and 954 mg/dL) also were negative for urine ketones but did exhibit blood levels of β-hydroxybutyrate of 10 and 20 mg/dL. Ketone bodies in urine appear only after the establishment of pronounced ketonemia. It should be noted that ketone bodies in the blood of chronically diabetic mice are considerably lower than those in humans, probably because mice can metabolize blood ketones to lactate (20). As hyperglycemia and glycosuria become more severe, mice exhibit polyuria, dehydration, and weight loss. Complete protocols for monitoring the diabetic state have been published elsewhere (18). In vivaria where high levels of NOD intra-islet insulitis develop, but unknown environmental factors suppress the transition to clinical type 1 diabetes, the onset of disease in both males and females can be precipitated by treatment with cyclophosphamide (21). Commonly, 200 mg/kg of body weight is administered intraperitoneally initially, with a second injection given 2 weeks later if required. Cyclophosphamide-accelerated type 1 diabetes is a particularly useful tool for synchronizing activation of diabetogenic effectors in young (7- to 10-week-old) NOD/Lt males.

It is exceedingly difficult to maintain stable normoglycemia in a diabetic mouse (NOD or otherwise) by insulin injection, although life span can be extended. A protocol used by Dr. M. Hattori (Joslin Diabetes Center, Boston, MA) (22) entails twice-daily subcutaneous insulin (a 1:1 mixture of regular and NPH insulin administered at 7:30 a.m. and 4:00 p.m.). The dose of insulin (between 0.5 and 1.0 units) is adjusted according to the severity of matinal and evening glycosuria. Table 18.2 shows random glucose and glycosylated hemoglobin (HbA₁) measurements in diabetic animals with and without insulin treatment.

An important issue for investigators studying genes and phenotypes in the NOD mouse is that of an appropriate control. The literature shows that C57BL/6J (B6) or BALB/c are commonly used as “standard” strains against which to compare NOD immunophenotypes or gene sequences. However, since neither strain was Swiss-derived, the differences observed could represent normal variation expected for independently derived mouse strains. Selection of a control strain depends on the nature of the experiment and the hypotheses being tested. A partial listing of possible control strains and their potential uses is provided in Table 18.3. For an investigator interested in associating a particular genetic polymorphism with a phenotype associated with NOD diabetes, it is logical to compare the NOD genotype/phenotype with that found in NOD-related but type 1 diabetes-resistant inbred strains. These include mice with a common Swiss origin but with differences in their major histocompatiblity complex (MHC) haplotype and/or other non-MHC genes (designated Idd loci) conferring various degrees of resistance to type 1 diabetes. Figure 18.1 provides an illustration wherein Swiss-derived strains were surveyed for the phenotype of high percentages of T lymphocytes in the peripheral blood and spleen, an NOD strain characteristic very possibly associated with the resistance of these T lymphocytes to apoptosis (23). This complex phenotype is not unique to NOD; it also is characteristic of the SJL/J strain. SJL/J (H2s) mice are highly susceptible to experimental allergic encephalomyelitis (EAE). A genetic contribution to the SJL strain susceptibility to EAE colocalizes to a 0.15-cM region on chromosome (Chr) 3 that also confers susceptibility to type 1 diabetes (Idd3) in outcrosses of NOD with certain type 1 diabetes-resistant strains (24,25). The logical candidate within this interval is the Il2 gene, with NOD and SJL showing identity for an Il2 gene polymorphism associated with a presumed hyperglycosylated interleukin-2 (IL-2) molecule (26,27). However, it is unknown whether the SJL trait of high numbers of peripheral T cells cosegregates with EAE susceptibility and the SJL Il2 allele. Indeed, SWR/J, another Swiss-derived strain, also shares the Il2 polymorphism with NOD but does not exhibit the T lymphoaccumulation phenotype (Fig. 18.1). This does not exclude the Il2 polymorphism as a contributor to the phenotype but illustrates that genetic control of the trait is oligo- or multigenic. These examples also illustrate the value of comparing NOD to more than one control strain.

Because it is H2^g7 -identical, but diabetes- and insulitis-resistant, NOR/Lt is an NOD-related strain frequently used to evaluate the effect of changes in non-MHC genes. This is a recombinant congenic strain arising from outcross of NOD with
C57BLKS/J (28). As discussed in detail below, the major genetic determinant of diabetes susceptibility in NOD mice is their H2^g7 MHC haplotype (K^d, A^g7, E^null, D^b). If an investigator is interested in analyzing the diabetogenic functions of the NOD H2^g7 MHC haplotype, stocks congenic for MHC haplotypes conferring resistance to type 1 diabetes are available for analysis of the selection of an autoimmune repertoire. Conversely, the diabetogenic NOD H2^g7 haplotype is available for study on the type 1 diabetes-resistant B6 background and in recombinant congenic strains (RCS) between CBA and NOD (29). During the initial period of NOD investigations in Japan, destruction of pancreatic β-cells was firmly established as a T lymphocyte-mediated process [reviewed in reference (10)]. For analysis of the diabetogenic potency of specific NOD T-cell clones or populations of immune effectors, NOD stocks congenic for immunodeficiency-producing mutations blocking development of functional T lymphocytes and/or B lymphocytes are available for adoptive-transfer studies. It has recently been demonstrated that B lymphocytes are essential antigen-presenting cells (APCs) for initiation of type 1 diabetes (B lymphocyte-deficient NOD mice congenic for the Igμ targeted mutation were protected from type 1 diabetes) (30). Therefore, T and B lymphocyte-deficient NOD mice congenic for either the Prkdc^scid (severe combined immunodeficiency mutation; henceforth denoted as scid) or targeted mutations in the recombinase-activating gene (Rag1, Rag2) provide useful diabetes- and insulitis-free mice for adoptive-transfer studies without a requirement for prior irradiation (31,32,33,34). The NOD- Rag1 and – Rag2 congenics do not exhibit the DNA repair defects associated with the scid mutation and hence can tolerate sublethal irradiation. Moreover, thymic lymphomagenesis, a strain characteristic of NOD- scid, is slower to develop in NOD- Rag mice (32). Further, these stocks provide an excellent source of NOD pancreatic islets free of insulitic infiltrates. Not included in Table 18.3 is a listing of a growing battery of congenic stocks with defined chromosomal regions bearing non-MHC Idd loci affecting diabetes-related immunophenotypes (35). Certain of these will be discussed below in the section on genetic control of type 1 diabetes.

Insulitis, the disruption of islet structure and function by infiltrating leukocytes, is the major pathognomonic feature of the development of type 1 diabetes in NOD mice (8). In these mice, the development of insulitis is not abrupt; rather, the first stages are detectable in females around the time of weaning (4 weeks), at least 2 to 3 months before development of overt diabetes. Islets develop in the perivascular/periductular areas of the pancreas. Insulitis initiates as “periinsulitis,” a pervasive leukocytic infiltrate emanating from the pancreatic vasculature and secretory ducts. Periinsulitis is not unique to NOD/Lt mice; for example, it can be detected in older mice of the related, but diabetes-resistant, NON/Lt strain, as well as in the diabetes-resistant NOR/Lt strain (36,37). However, what distinguishes the histologic appearance of insulitis in NOD mice from that observed either in the T-lymphopenic BBDR (BB diabetes-prone) rat or in humans is the unusually large numbers of leukocytes associated with all the pancreatic vascular spaces, including the lymphatics (lymphoaccumulation) (38). Lymphoaccumulation is not restricted to the pancreas; increased percentages of T lymphocytes are found in peripheral blood, in lymphoid organs, and in other endocrine tissues. Data in Figure 18.1 comparing NOD/Lt to other ICR (Swiss)-derived inbred strains illustrate this phenomenon. Perhaps reflective of this phenomenon, lymphoid tumors are common in aging NOD/Lt mice that remain free of type 1 diabetes (39). Accumulation of leukocytic infiltrates around the perivascular and periductular regions of the pancreas and subsequent development of progressively more severe insulitis (Fig. 18.2), usually initiates between 3 and 5 weeks in NOD females and several weeks later in males. These intrapancreatic lymphoaccumulations primarily, but not exclusively, comprise T lymphocytes (CD4⁺>CD8⁺) (40,41,42). B lymphocytes and macrophage/dendritic cells also are found in the infiltrates (43). Although initially few in number, T lymphocytes capable of adoptively transferring type 1 diabetes into T and B lymphocyte-deficient NOD/LtSz- scid recipients can be isolated from islet donors as young as 3 weeks old (44).

Periinsular leukocytic aggregates around NOD islets usually initiate at one pole but eventually surround the entire islet perimeter (periinsulitis). After initiation of periinsulitis (usually in 3- to 5-week-old females and several weeks later in males), only a subset of islets observed in a microscopic field appears to be penetrated by leukocytes. However, as the mice transit through puberty, an increased prevalence of islet profiles showing intra-islet leukocytic infiltrates, coupled with erosion of between 25% and 90% of the insulin-stainable β-cell mass, is noted (18). The early insulitis, although leading to complete destruction of a percentage of islets in a given pancreas, appears
to be partially compensated by a period of islet growth. During the period between 5 and 12 weeks, NOD/Lt islets are actually quite large in comparison to those of the closely related NON/Lt strain despite the heavy leukocytic aggregations surrounding or penetrating many of the islets in the former. Basal plasma insulin levels are significantly higher in NOD mice of both sexes between 4 and 6 weeks of age compared with levels in C57BL/6 (B6) mice bred in the same vivarium (45). Similarly, NOD fetal pancreas in organ culture secretes higher levels of insulin than does fetal BALB/c pancreas (46). Marked decreases in pancreatic insulin content are demonstrable in NOD/Lt females at around 12 weeks of age and several weeks later in males (47). At this time, increased numbers of atrophic “end-stage” islet profiles are detected. Immunocytochemical staining of these islet residua detects primarily non-β-islet endocrine cells. When more than 90% of the β-cell mass of the pancreas is destroyed, clinical symptoms of diabetes (hyperglycemia, glycosuria, polydipsia, and polyuria) appear abruptly (48). Although intra-islet insulitis initiates early after weaning, some immunoregulatory mechanism apparently keeps many of the infiltrating T lymphocytes in a resting state until a later activation event allows more widespread destruction of the β-cell mass (47,49).

In considering the insulitis process, the reader should be reminded that important substrain and environmental differences distinguish extant colonies of NOD mice around the world (39). Hence, a longitudinal analysis of cellular and molecular events descriptive of insulitic progression in one colony may not exactly match an analysis of the same parameters in a separate colony. This is especially true for reverse transcriptase-polymerase chain reaction (RT-PCR) estimations of cytokine messenger RNAs (mRNAs) expressed in longitudinal studies.

Islet-associated macrophages may mediate cytopathic events indirectly (through secretion of lymphocyte chemoattractants and through the presentation of autoantigens to them) or directly (via secretion of toxic monokines, prostanoids, nitric oxide, and other reactive oxygen species). An immunocytochemical study of NOD mice from a colony in Paris detected ER-MP23⁺ and MOMA-1⁺ macrophages/dendritic-like cells as
the first leukocytes to appear around swollen periinsular vessels in 3-week-old mice of both sexes (50); this was termed “macrophage insulitis.” At 7 weeks, periinsular accumulations of leukocytes were reported that contained, in addition to the ER-MP23⁺ and MOMA-1⁺ cells, a BM8⁺ cell described as a phagocytotic macrophage. A shift in location from the periinsular space to the islet interior of BM8⁺ macrophages was associated with intra-islet insulitis and occurred earlier and at higher frequency in the pancreas of females than in the pancreas of males (50). In NOR/Lt mice, in which heavy periinsulitis develops with age, macrophages penetrate into the islet interior but T cells do not, possibly suggesting absence of a critical chemokine (36). Analysis by RT-PCR of spontaneously developing insulitis in NOD mice shows that, although individual islets within a single pancreas may exhibit a T-lymphocyte population expressing a relatively oligoclonal spectrum of T-cell receptor (TCR) genes, pooled islets exhibit a much more diverse spectrum of TCR clonotypes (51). The islet-reactive CD4⁺ T-cell clones now available from NOD spleens or islets show a diverse array of TCR α- and β-chain rearrangements (52,53). Interestingly, among the H2-K^d-restricted CD8⁺ islet-reactive clones isolated from NOD pancreatic islets, a more restricted TCR-α gene utilization has been noted, with up to 30% to 35% using an identical Vα17 to Jα42 TCR gene rearrangement (54,55).

As discussed in more detail below, CD4⁺ T lymphocytes of NOD mice are strongly biased to secrete a T-helper 1 (T_H1) spectrum of proinflammatory lymphokines, especially interferon-γ (IFN-γ). IL-12 and IL-18 are two cytokines associated with the deviation of T lymphocytes toward a T_H1 cytokine profile. Both are expressed at high levels in NOD splenic and islet-infiltrating macrophages (56,57). In the cyclophosphamide (CY)-accelerated model of diabetogenesis in NOD mice, a T_H1-insulitis in NOD/Bom was correlated with upregulation of IL-12 and IL-18 mRNA levels in both spleen and pancreas (56,57). The importance of these monokines is indicated by the observation that onset of type 1 diabetes was retarded by a T_H2 deviation achieved by chronic treatment of prediabetic NOD/Lt females with an IL-12 antagonist (58). An elegant quantitative RT-PCR analysis found I-Aβ^g7 and Igμ transcript levels to be elevated by 20-fold in islets from 20-day-old NOD/Jsd mice compared with nondiabetic control strains (59). However, these markers for intra-islet APCs encompassing both macrophages and B lymphocytes failed to distinguish NOD/Jsd females that transited to overt type 1 diabetes at a high frequency (80% to 85% by 30 weeks) from NOD/Jsd males that were quite resistant to type 1 diabetes (10% to 15% diabetic by 30 weeks). Fox and Danska (36) subsequently used their quantitative RT-PCR techniques to compare a variety of APC-specific and T lymphocyte-specific transcript levels in NOD/Jsd versus the H2^g7 identical, but insulitis- and diabetes-resistant NORLt strain. They reported independent genetic regulation of the macrophage insulitis (found in NOD and NOR) from the more advanced phase involving T-lymphocyte penetration (found primarily in NOD) (36). These findings support the concept that clear “checkpoints” in the development of insulitis exist. The swollen periinsular vasculature prior to lymphocyte penetration of islets described above is associated with increased expression of MHC class II and addressin on vascular endothelium. These changes probably facilitate the NOD strain-characteristic T-lymphocyte accumulation in the pancreatic lymphatics and the subsequent extravasation of α₄β₇-integrin-positive lymphocytes into the periinsular zones (60,61).

Small numbers of T lymphocytes (both CD4⁺ and CD8⁺), as well as B lymphocytes, are present in the early insulitic lesions (41,62). Studies of adoptive transfer into NODLtSz- scid mice confirmed previous reports that pathogenic contributions from both CD4⁺ and CD8⁺ subsets are required in the natural progression of the disease (63,64). CD4⁺ splenic T lymphocytes preactivated in situ in overtly diabetic donors can adoptively transfer type 1 diabetes into NOD/LtSz- scid recipients in the absence of the CD8⁺ subset. In contrast, both CD4⁺ and CD8⁺ subsets are required to transfer type 1 diabetes if isolated from young, prediabetic donors (63). In NOD/LtSz- scid mice, CD8⁺ T lymphocytes adoptively transferred into NOD- scid recipients cannot home to islets in the absence of CD4⁺ cells (63). The mutual dependency of CD4⁺ and C8⁺ T lymphocytes for initiation of disease in adoptive transfers into NOD- scid is elegantly illustrated by using NOD- scid recipients also homozygous for a targeted β-2 microglobulin (B2m) gene. The islets in these recipients do not express MHC class I cell surface molecules and thus do not serve as CD8⁺ targets. When splenic T lymphocytes from NOD donors of various ages are transferred into this immunodeficient stock, only those from overtly diabetic donors transfer type 1 diabetes in the absence of class I-expressing islets (54,65). Thus, the earliest initiation, and all but the final phases of autoimmune pancreatic β-cell destruction, require MHC class I-dependent T cells that are able to recognize cognate antigen on MHC class I-expressing β-cell targets. The contributions of CD4⁺ T cell-secreted cytokines to the development of insulitis through various checkpoints are discussed in the next section.

Adoptive-transfer studies have focused attention on the pathogenic contributions of T lymphocytes, which, when derived from diabetic donors, are able to transfer disease into young recipients without recruitment of host B lymphocytes (66). However, B lymphocytes serving as APCs to amplify T-lymphocyte responses to β-cell autoantigens are essential diabetogenic catalysts in NOD mice. This was established by the discovery that NOD/Lt mice rendered B lymphocyte-deficient either by congenic transfer of a disrupted immunoglobulin heavy-chain gene (Igμ “knockout”) or by treatment with antibody to μ chain rarely developed type 1 diabetes (67,68,69). The absence of B lymphocytes rather than the potential presence of any diabetes resistance genes inadvertently transferred with the genetically disrupted Igμ locus on Chr 12 was demonstrated by abrogation of resistance to type 1 diabetes following repopulation of these mice with B lymphocytes. Autoantibodies, most likely maternal in origin, are found on β-cells from islets of 2- to 3-week-old donors (41). However, such autoantibodies are apparently not required for pathogenesis, since passive transfer of purified serum immunoglobulins from diabetic NOD/Lt donors into these NOD/Lt. Igμ^null mice, in contrast to repopulation by B lymphocytes, failed to abrogate the resistance to type 1 diabetes of this stock. It is not known whether the B lymphocytes infiltrating the islet play essential roles in presentation to T lymphocytes of β-cell autoantigens released by T lymphocyte- and/or macrophage-catalyzed β-cell lysis.

One of the most unusual aspects of the diabetogenic process in NOD mice is that the autoimmune destruction of β-cells can be drastically retarded by so many different manipulations (70). In essence, the immune dysregulation of NOD mice that predisposes the strain to type 1 diabetes, particularly impaired APC function, can be partially corrected by immunostimulation (71). In some instances, the damage mediated by the insulitic process can be halted or reversed at quite a late stage in disease progression (72). In almost all cases in which the consequences of a protective manipulation have been investigated at the level of
cytokine expression in the pancreas and, more germane, in the islet-infiltrating lymphocytes, a deviation in the spectrum of cytokine profiles expressed is indicated that suggests a T_H1→T_H2 shift has occurred (73,74). Whereas more than 140 manipulations can drastically retard the onset of hyperglycemia in NOD mice, a much smaller number (mostly entailing transgenic or gene targeting manipulations of genes associated with antigen presentation) are capable of completely suppressing the development of insulitis. This paradox of disease suppression, but not insulitis elimination, has given rise to the concept that a “benign” or “nondestructive” form of insulitis that is present in juvenile NOD mice (or NOD male mice in some colonies) can either be maintained or be elicited through maturity. It is widely held that T lymphocytes producing proinflammatory cytokines associated with the T_H1 phenotype, especially IFN-γ, are more likely to effect tissue damage in either spontaneous or experimentally induced T lymphocyte-mediated diseases in mice. Although IL-2 is sometimes used as a paradigm T_H1 cytokine, NOD/Lt T lymphocytes are low producers of IL-2 on a per-cell basis, and the NOD’s IL-2 allele may prove to be the susceptibility gene at the Idd3 locus (75). T_H2 functions [especially the production of IL-4, IL-10, and transforming growth factor-β (TGF-β)] are viewed as protective, or at least as neutral. The genomic makeup of NOD mice predisposes their CD4⁺ T lymphocytes to respond to antigenic stimulation in a T_H1-biased fashion. As noted previously, NOD macrophages express high levels of IL-12 and IL-18 mRNA (56,57,76,77). This would perhaps partially explain why IFN-γ responses dominate over IL-4/IL-10 secretory responses in NOD mice.

A longitudinal analysis of cytokine gene expression in a colony of NOD/Jsd mice with a marked gender difference in the frequency of type 1 diabetes illustrates the changing cytokine profile that accompanies the progression of NOD mice toward frank hyperglycemia (59). Quantitative RT-PCR analysis showed that islet-infiltrating lymphocytes in males maintained a higher ratio of IL-4/IFN-γ transcripts through puberty than did females, with the increased likelihood of the development of type 1 diabetes in the latter associated with waning expression of T_H2 transcripts (59). These types of data reinforce the concept that insulitic destruction of β-cells in NOD mice is represented neither by a continuous linear regression line nor by an abrupt, late-activating process but rather by a gradual and chaotic progression wherein both regulatory events (including the changing endocrine milieu imposed by puberty and reproduction) and stochastic events (the particular subpopulation of CD4⁺ and CD8⁺ T lymphocytes “resident” in any given islet) determine the rate at which the diabetes threshold of more than 90% β-cell destruction is reached. The finding that many immunomodulatory interventions that retard type 1 diabetes development in this model produce a T_H1→T_H2 deviation is generally assumed to be evidence that the deviation itself was the basis for the retarded destruction of β-cells by islet-infiltrating effectors. However, recent analysis of NOD stocks congenic for disrupted genes encoding IFN-γ, IFN-γ receptor β chain, IL-4, and IL-10 has forced a reevaluation of this rather simplistic, but attractive concept.

NOD mice congenic for either a disrupted IFN-γ or IFN-γ receptor β gene (e.g., unresponsive to IFN-γ signaling while producing high levels of T_H2 cytokines) develop type 1 diabetes, while NOD stocks homozygous for disrupted IL-4 and IL-10 genes do not become diabetic at an accelerated rate (35,78,79). Although NOD mice congenic for a disrupted IFN-γ receptor α subunit have been reported to be resistant to type 1 diabetes, a subsequent report indicates that the resistance is not the result of the targeted allele on Chr 10 but rather a contribution of strain 129 genome in tight linkage (80,81). Treatment of standard prediabetic NOD mice with the nonspecific immunomodulator BCG (bacille Calmette-Guérin) induced resistance to type 1 diabetes resistance with a T_H1→T_H2 deviation. This treatment also protected NOD stocks congenic for disrupted IL-4 or IL-10 genes (35,82). Surprisingly, the IFN-γ-deficient stock is not protected by BCG (35). Collectively, these observations suggest that the T_H1→T_H2 deviation so commonly associated with protective immunomodulatory protocols are consequences, rather than causes, of the deviation. A number of studies have shown that NOD APCs, particularly macrophages, are not fully mature and hence fail to provide normal costimulatory signals such as IL-1, while producing high levels of prostaglandin E₂ (83,84). The resistance of NOD peripheral T lymphocytes to activation-induced cell death (AICD) may explain the T_H1 bias, since T_H1 cells normally are more prone to AICD than are T_H2 cells (85,86). Hence, if certain type 1 diabetes-preventative protocols differentially stimulate the AICD of β-cell autoreactive T lymphocytes in a way that preferentially spares those producing T_H2 cytokines, it would give the appearance of a T_H1→T_H2 cytokine shift (35).

A number of genetic and molecular genetic manipulations designed to limit the T-cell receptor repertoire available to CD4⁺ and/or CD8⁺ T cells in the NOD mouse generally fail to prevent type 1 diabetes (87). Hence, the NOD thymus is extremely flexible in its ability to generate a diabetogenic T-cell repertoire, even when rearranged TCR transgenes dictate a repertoire predominantly skewed to an “irrelevant” antigen [e.g., lymphocytic choriomeningitis virus (LCMV) peptide not expressed in NOD β-cells]. The broad spectrum of TCR clonotypes of both CD4⁺ and CD8⁺ T cell lines and clones reported to produce β-cytotoxicity effects in the NOD mouse reinforce the diversity of antigens recognized. An unresolved issue is whether a single β-cell autoantigen is targeted initially, followed by “downstream” immune reactivities as other antigens are processed and presented following the initial destruction of a few β-cells. The most commonly discussed candidate autoantigens in humans (based upon spontaneous autoantibody development) are insulin, glutamic acid decarboxylase (GAD), and IA-2, a putative tyrosine phosphatase localized in β-granules (88,89,90,91). Debate about NOD mice has focused on whether insulin or GAD represents the primary autoantigen. T-lymphocyte responses to GAD peptides appear earlier than to other candidate autoantigens, including carboxypeptidase E, peripherin, and heat-shock protein 60 (92). Administration of autoantigens in tolerogenic doses to prediabetic NOD mice can retard the development of type 1 diabetes (93,94). A rather controversial report that GAD67 is the primary autoantigen in NOD β-cells has not been replicated and is suspect because of the claims of easily detectable GAD67 protein in NOD β-cells by immunocytochemistry (95). Based on the plasticity of the T-cell repertoire capable of destroying NOD islets, the most conservative position regarding β-cell autoantigens in the NOD model is that a multiplicity must exist, each individually capable of serving as a target for autoimmune initiation. As noted above, CD8⁺ T lymphocytes are essential for such initiation, suggesting that the initiating event entails recognition of autoantigenic peptide presented by the β-cell target or vascular endothelium near the β-cells.

Most studies examining the interaction between NOD T lymphocytes and their β-cell targets assume that if the appropriate autoantigen(s) is present and there are T lymphocytes in the periphery with the appropriate TCR clonotype(s), then type
1 diabetes will ensue. This is not necessarily true. Both rodent and human β-cells have very weak defenses against oxidative stress (96). NOD β-cells appear typical of most mouse strains in being quite sensitive to cell death mediated by combinations of monokines and IFN-γ. Analysis of β-cells from ALR/Lt mice, a strain closely related to NOD and selected for resistance to type 1 diabetes induced by a low dose of alloxan, has demonstrated that genes expressed at the β-cell level, and presumably associated with dissipation of reactive oxygen species, can protect against both toxic combinations of cytokines and monokines, as well as against islet-reactive cytotoxic T cells isolated from NOD islets and capable of recognizing antigens presented by ALR/Lt islets (13,97,98,99).

NOD mice are susceptible to a variety of experimentally induced organ-specific immune diseases, including drug-induced diabetes, lupus, colitis, encephalomyelitis, and thyroiditis. Not surprisingly, young NOD/Lt males, like outbred CD1 males, are susceptible to induction of type 1 diabetes by multiple low doses of the fungal antibiotic streptozotocin (190). This sensitivity is independent of autoimmune T-cell involvement,
since NOD/LtSz- scid males are even more sensitive to low doses of this diabetogen (190). Injection of a single dose of 2.6 × 10⁷ heat-killed Mycobacterium tuberculosis (BCG) intravenously into 8-week-old NOD mice prevented diabetes but precipitated a syndrome similar to systemic lupus erythematosus (191). Exposure of NOD/Lt mice to 3.5% dextran sodium sulfate in the drinking water for 5 days elicits a severe colitis (192). Experimental allergic encephalomyelitis can be induced in NOD mice using a peptide fragment (residues 56–70) in adjuvant from proteolipid protein (193). The NOD allele at the “Idd3” locus (in the region of the gene encoding IL-2) has been shown to be a component of this strain’s high susceptibility to encephalitis induction (24). A similar demyelinating disease can be induced by injection of the MOG peptide (residues 35–55) in adjuvant; this model is prevented by treatment with anti-IL-12 antibody. A single intravenous injection of pertussin without adjuvant into NOD mice induced a T-cell-mediated encephalitis (194).

BioBreeding (BB) rats are a diabetes-prone strain derived from outbred Wistar rats from a commercial colony in Canada (195). The interested reader is referred to comprehensive recent reviews (196,197,198,199,200). Spontaneous type 1 diabetes was discovered in 1974, and selective breeding for that phenotype ensued. As with NOD mice, multiple substrains of BB rats exist, and lymphoid infiltrates are not limited to the pancreatic islets (201). The best-characterized BB colonies are those developed at the University of Massachusetts Medical School and now commercially distributed by Biomedical Research Models (Worcester, MA). A diabetes-resistant substrain (BBDR/Wor) has been selected, as have several diabetes-prone (BBDP/Wor) lines. These BB rats are fully inbred and characterized. In the Worcester virus-antibody-free (VAF) barrier colony, the frequency of type 1 diabetes in both sexes of DP rats exceeds 90%, whereas no spontaneous type 1 diabetes develops in VAF BBDR/Wor rats. A major genetic and phenotypic difference distinguishing the DR from the various DP strains is the presence in the DP line of a recessive mutation, lymphopenia (Lyp), on Chr 4 (202,203,204). Although DP rats develop a thymus of normal size and cellularity, recent thymic T lymphocyte emigrants to the periphery [phenotypically Thy-1⁺ and not yet expressing a cell-surface adenosine diphosphate (ADP)-ribosyltransferase (formerly designated RT6 and now designated ART2)] are very short-lived (205,206). The consequence is a profound T-lymphopenia/lymphocytopenia. Hence, these rats are severely immunocompromised, as reflected by impaired T-lymphocyte function in vitro, impaired ability to reject foreign tissue grafts, increased susceptibility to B-lymphomagenesis, and requirement for special husbandry procedures to protect against infections by opportunistic pathogens. This obviously contrasts sharply with NOD mice, whose T lymphocytes are long-lived in the periphery, present in large numbers, and express a functional ortholog of rat RT6 (now designated as ART2 in both genera) (207). Whereas BBDP rats are strongly immunocompromised, as evidenced by difficulty in rejecting non-islet allografted tissues and being susceptible to infectious pathogens, NOD T cells are capable of rapidly rejecting allografted tissues, and NOD mice are remarkably pathogen-resistant. There are many additional distinctions between BB rats and NOD mice in the diabetes syndrome itself, reviewed elsewhere, and some are summarized in Table 18.4 (6,197,198).

Superficially, the genetic basis for the spontaneous development of type 1 diabetes in rat models appears less complex than that in the NOD mouse or in humans. However, this may relate to the level of study of the process, since initially the genetics of type 1 diabetes in NOD mice was also inferred to be less complex, and possibly only oligogenic (100,102). As in humans and mice, the detection of “Idd” loci in the rat genome has been facilitated by the development of polymorphic microsatellite markers that can now be scored by automated methods (201). Segregation analysis between BBDP crossed with other type 1 diabetes-resistant strains have established that the MHC class II alleles encoded by the RT1^u haplotype of the BB rat are essential for diabetogenesis (201,208,209,210,211). This MHC susceptibility, designated “Iddm2,” is permissive for insulitis in the heterozygous state; however, most diabetic segregants are MHC homozygotes. Analyses of insulitis and development of type 1 diabetes following outcross of diabetes-prone BB/Wor rats of the “NB” subline to RT1^u congenic stocks of PVG rats carrying recombinants in the class I versus class II region have established the requirement for the RT1-D^u/B^u alleles at the two class II loci (homologues of human DQ and DR, respectively) (210). Unlike
the nonexpression of H2-E molecules on APC of the NOD mouse, APC of the BB rat do express the DR homologue (RT1-D^u). Sequence analysis of the BB rat DQ homologue (RT1-B^u) showed it differed from both the high diabetes risk-conferring human HLA-DQβ alleles and the NOD H2-Ab^g7 molecule. In contrast to NOD mice and humans, the BB rat allele contains an aspartic acid at residue 57 of the β chain instead of exhibiting “diabetogenic” non-Asp substitutions (212,213). All RT1^u-expressing rat strains carry the requisite class II “Iddm2” alleles. Class I alleles from non-RT1^u-positive rat strains apparently will substitute for the RT1-A^u, E^u, and C^u class I alleles. Interestingly, although insulitis and thyroiditis were suppressed in class II heterozygous segregants in one outcross/backcross analysis, thyroiditis (another characteristic immune pathology of the BBDP-NB subline used in the outcross) was not (210).