Autoreactivity, by definition, designates a specific immune response to self-antigens. Antigen nonspecific responses such as inflammatory and innate immune processes should not be considered autoimmune in the strict sense, although they may accompany, enhance, or even trigger autoimmune processes proper. Thus, antigen-specific T- or B-cell immunity will have to underlie a genuine autoimmune disorder. Furthermore, for organ-specific autoimmune diseases, antigen specificity of primary effector lymphocytes must be largely restricted to autoantigens derived from defined organs or tissues. Once initiated, specific responses that precipitate or “drive” the localized autoimmune reaction may diversify to comprise additional specificities (determinant spreading) and pathogenic mechanisms.

How does the adaptive immune system restrict generation and activation of autoreactive lymphocytes? The central process by which the generation of TCR diversity is limited is called thymic selection. Thymic selection is a developmental process that selects T cells with a biased repertoire for export into the periphery.^{75,76,77,78,79} T cells that interact at least weakly with self-peptides presented in the context of MHC molecules are chosen in the course of positive selection,^80,81 while those that do not effectively interact with MHC/peptide complexes die “by neglect.” However, interactions above a certain avidity threshold result in elimination by negative selection and constitute the basis for “central tolerance.”^75,79 Thus, central tolerance prevents widespread autoimmunity as a function of lymphocyte/antigen-complex avidity and preferentially selects T cells with specificity for antigens not expressed in thymic epithelium for export into the periphery. However, central tolerance is not a complete mechanism, and a sizable pool of T cells with intermediate avidity can escape negative selection and constitutes most autoreactive T cells found in the peripheral immune system. The presence of these autoreactive T cells can be considered physiologic, and they are not noxious for two reasons: either they are usually not activated and exhibit a “naive” phenotype or they exhibit regulatory effector functions and act as adaptive T_regs following activation. Thus, not all self-reactive lymphocytes need to necessarily exhibit an aggressive phenotype. Depending on their specific effector functions, autoreactive T cells may exhibit regulatory functions and may critically modulate or even abort local autoimmune processes. Such autoreactive regulators might occur physiologically and constitute most autoimmune responses present in healthy individuals. Only an encounter under appropriate stimulatory conditions (ie, presentation of autoantigen-derived peptides presented in the context of MHC class I or II molecules accompanied by antigen-nonspecific costimulatory interactions and strong inflammatory signals) can lead to their full activation and detrimental effector functions in the periphery. As such “armed effectors,” autoreactive T cells are now potentially very dangerous and may initiate specific autoimmunity, if they recognize the autoantigens or closely related proteins in a defined tissue. It is thought that a few autoaggressive “driver clones” with highly detrimental effector function can sustain a localized autoimmune process. High receptor/MHC/self-peptide avidity likely, but not necessarily, predisposes to this phenotype.^82,83,84

The presence of autoreactive T cells in the periphery might suggest that detrimental autoimmunity should occur quite frequently if organ-specific autoantigens are not expressed in the thymus, or alternatively or in addition, such physiologically occurring autoimmunity is not of a regulatory nature. Yet, there are several additional mechanisms that maintain tolerance in the periphery. “Peripheral tolerance” involves a set of mechanisms that ensures that autoreactive T lymphocytes are not activated in the periphery. It should be noted that these mechanisms pertain to both autoreactive and “heteroreactive” T cells and involve the following pathways. First, it has been observed that naive T cells triggered by a strong signal through the TCR alone may lose the ability to proliferate, and some but not all effector functions become “anergic.”^85,86,87 Presence of certain cytokines or costimulatory interactions can avoid the induction of anergy or may reverse an anergic state. Second, highly activated T cells will eventually undergo activationinduced cell death (AICD).⁸⁸ AICD is thought to be essential for the downmodulation of immune responses and the reestablishment of immune homeostasis. Impairment of AICD may lead to continued immune activation and generalized autoreactivity. For CD4 lymphocytes, AICD is FAS/FAS-L dependent^89,90; it is not clear which interactions precisely control AICD in CD8 cells. Third, molecules that can deliver specific negative signals, such as the B7-binding cytotoxic T-lymphocyte antigen (CTLA)-4 are involved in “turning off” of antigen-specific T cells.^91,92 Finally, other factors such as regulatory lymphocytes and regulatory antigen-presenting cells (APCs) might play important roles in maintaining peripheral tolerance.^59,70,93

As indicated previously, while autoimmune disorders must conform with a set of general criteria, organ-specific autoimmunity must be considered in the context of the target organ affected. Certain effector functions exerted by autoreactive lymphocytes will be detrimental only to particular cells or tissues. For example, the pancreatic β cells are more sensitive to the damaging effect of inflammatory cytokines than neighboring α cells (both cell types are part of the islets of Langerhans) and other cells in their vicinity (eg, fibroblasts or acinar tissue of the pancreas).⁹⁴ On the other hand, some organs provide a microenvironment that suppresses inflammatory responses. For example, both the gut and the central nervous system (CNS) contain relatively large amounts of TGF-β, which can have direct anti-inflammatory effector functions on T cells and APCs, and activate T_regs, unless IL-6 is present.^{69,95,96,97,98} Also, certain large organs such as the liver may better tolerize lymphocyte responses directed toward them because they contain a high number of cells that are incapable of costimulation (eg, hepatocytes) and will, therefore, more likely shut down naive autoreactive lymphocytes that stimulate them.³⁰ Finally, the precise activation state, phenotype, and effector function of an autoreactive cell is critical in determining their impact on tissues expressing their cognate antigen. Some molecules will exert beneficial functions, for example, IL-4 and probably IL-10 and TGF-β in T1D.^99,100,101 Antigen-specific adaptive and professional regulatory cells are likely participants of every local autoimmune process and can therefore be essential for delay or prevention of clinical disease altogether. While their induction is a clearly therapeutic goal, phenotype and mechanistic aspects essential for regulatory function are not yet understood in complete detail. Likely, professional and adaptive T_regs may exert their function by targeted alteration of APC function. Recent evidence even suggests that T_regs may safeguard against autoimmunity by specifically killing APCs that present autoantigens to effector T cells.¹⁰² They may also act as true bystander processors if their suppressive action via APC modulation is extended to aggressive immune responses regardless of specificity found in the microenvironment of the affected organ (Fig. 44.2). Indeed, such cells have been found in several animal models,^{100,103,104,105} but their existence and function in humans is not well defined. Alternatively, regulatory lymphocytes may directly affect activated or naive aggressive lymphocytes and induce anergy or apoptosis. Lastly, by changing the overall cytokine milieu of a given inflammatory process, the number and function of aggressive cells and cytokines/chemokines may be dampened in a localized area.¹⁰⁶ Probably, the best understood balancing circuit employed by autoreactive adaptive T_regs are the T_H1/T_H2 paradigm^107,108,109 and signaling by IL-10⁶⁸ and/or TGF-β.⁶⁹ In light of these observations, autoreactivity is not necessarily detrimental and maybe even essential in the sense of T_regs.

Last, it might be difficult or even impossible to define a general phenotype of autoaggressive lymphocytes that will cause damage at any site or organ. Rather, autoaggression has to be defined in relation to the target organ or cell that is under
attack, and lymphocytes detrimental in one organ/disease will not necessarily be detrimental to other organs, and ubiquitous autoimmunity will therefore occur very rarely.

A central challenge in most human autoimmune disorders is the identification of an initiating autoantigen or autoantigens. For example, several candidate autoantigens have been identified in T1D. However, to date, it remains unclear and controversial whether any are actually involved in initiation of the disease. Similar situations exist for multiple sclerosis and arthritis. Thus, distinguishing features that constitute an ideal candidate autoantigen and parameters by which potential candidates can be searched for and identified as the initiating autoantigen antigen for a given organ-specific disease must be determined with much greater precision to yield insights applicable to preventative or therapeutic interventions.

The concept that “processing of antigens determines the self” has been introduced by Sercarz.^110,111 Autoantigens that are not expressed in the thymus or are of cryptic nature^110,112,113 may encounter a more extensive T-cell repertoire that is likely of higher avidity. Such antigens appear to be better targets. Low levels of thymic antigen expression or limited numbers of thymic medullary cells may lead to partial tolerance.^114,115 In addition, the precise timing of antigen expression during embryonic development may play a major role for the establishment of tolerance. Antigens that are cryptic or not expressed during early embryonic development might become better targets later because more autoreactive T cells will be present in the periphery. It should be considered that certain “embryonic antigens” are involved in initiating autoimmune diseases.¹¹⁶

Of further importance are the constraints with which particular antigens can be presented to the immune system in a given target organ. Exogenous uptake of soluble antigens usually results in processing via the MHC class II presentation pathway. Such antigens may also be processed through the MHC class I pathway by dendritic cells, a mechanism termed cross-presentation, if the antigen itself was originally expressed by a cell different from the APC.^117,118 Cross-presentation appears to be less efficient than direct endogenous presentation of antigens through the MHC class I pathway. At this point, it is not clear which properties will make an antigen a more efficient candidate for cross-presentation. The class I pathway may be important in the effector phase for many autoimmune disorders because CD8+ T cells with cytotoxic function (cytotoxic T lymphocyte [CTL]) can be induced in this fashion, and CTLs have profound detrimental effector functions (interferon [IFN]γ and TNF-α secretion, lysis of target cells).¹¹⁹ Clearly, cross-presentation can lead to immunity, as well as tolerance, and in this way propagates or, alternatively, halts autoimmune disease.^117,120 It is less likely that autoantigens that are not presented by professional APCs or within lymphoid organs after their release from target cells/organs can initiate autoimmunity. Only under special circumstances (ie, extremely high-density presentation of an antigen on a target cell or highly inflammatory conditions including chemotaxis) would this appear possible. This might occur during localized or systemic viral infections and could also be an important event in lymphocyte entry into target organs, especially if chemotaxis is involved and endothelial cells present autoantigens.¹²¹ In addition, experienced or primed autoaggressive lymphocytes will be capable of becoming activated after seeing antigen in the absence of costimulation; thus, accumulating a memory pool of autoaggressive T cells might be inherently dangerous to an organism.

Thus, the type of autoantigen will define its own potential “candidacy” in an autoimmune process. Autoantigens
that are not expressed during development, or were cryptic for an extended period of time, as well as antigens secreted and cross-presented in lymphoid organs by APCs, appear to be better suited for assuming the role as primary culprits.

An abundance of empiric evidence indicates the association of many organ specific autoimmune diseases with certain human leukocyte antigen (HLA) haplotypes as well as other susceptibility or protective genes. The principal specific genetic linkages will be discussed separately for each condition in part II. In general, MHC class I (such as HLA-B27)^122,123 or class II (such as DR4) genes might predispose to a certain disease by enhanced presentation of pathogenic peptides in the periphery or inefficacious presentation of autoantigenderived peptides in the thymus. For example, the human HLA-DQ8 molecule has a striking structural similarity to mouse I-A^g7 class II that predisposes NOD mice to spontaneous autoimmune diabetes.^124,125 However, the link to disease appears not to be as simple as reasoned previously, and more complex mechanisms might be in place. Presentation of the pathogenic peptide in gluten-induced celiac disease involves acidic modification of the protein to generate the peptide ultimately presented by MHC class II,¹²⁶ and other chemical modifications can be expected to alter peptide binding to MHC and the resulting conformation of the peptide/MHC complex.

Other genes that encode immunoregulatory or inflammatory proteins may be involved in the disease process, and, finally, genes that support tissue or wound repair (eg, islet cell regeneration in T1D) may be of help in preventing disease development.¹²⁷ For most autoimmune disorders, the genetic links are complex, not absolute, and many susceptibility and resistance genes act in concert to modulate the clinical phenotype. Noteworthy, newly emerging themes in the genetics of autoimmunity are the fact that small nucleotide polymorphisms might result in alternate forms of the same gene and the fact that the character of the full array of MHC region genes is becoming understood.¹²⁸ A striking example for the former is the existence of an alternate spliced form of the inhibitor of cellular function, CTLA-4. Kuchroo and colleagues discovered that the presence of a secondary CTLA-4 lacking the B7.1/B7.2 binding motif can be instrumental in the genetic penetrance of diabetes pathogenesis.¹²⁹ An important area still largely unexplored is the nature of genes responsible for transcriptional control of other proteins. Experimental models indicate that transcription factors can have profound effects on the development of autoimmune disease, and variations in expression and activity levels may be found between susceptible and protected individuals. In summary, a complex interplay of many genes will predispose for a certain autoimmune disease, but the concordance of clinical manifestation is frequently not higher than 30% to 40% in monozygotic twins.^{130,131,132,133,134,135,136} For this reason, other factors are to be considered in triggering or propagating the pathogenic autoimmune process.

Our understanding of the genetic component to autoimmunity has greatly benefited from the arrival of a first wave of so-called genome-wide association studies (GWAS) in the past 5 years. Prior to the use of this technology, genetic contribution was analyzed by two different approaches: linkage analysis and association studies. Linkage analysis tests for correlation between a genomic region and phenotype in families, whereas association analysis searches for correlations in the entire population.¹³⁷ In a genetic association study of autoimmune disease, sequence variants are genotyped in patients, and controls and the frequencies of these genotypes are compared. It is then evaluated whether certain alleles are significantly more associated with patients than with controls. Most autoimmune conditions are polygenic diseases, meaning that they are caused by (a range of) combinations of polymorphisms within different genes. As mentioned previously, HLA-associated risk was discovered early in many autoimmune disorders and turned out to be reproducable because of its highly significant association. The vast majority of common genetic variants in autoimmunity, however, display modest effects. Even extremely large linkage studies have usually been unsuccessful in discovering susceptibility genes for polygenic diseases that met stringent statistical thresholds. Likewise, most common variants require large association studies for reliable detection. Advances in the development of high-throughput genotyping, statistical correlation methods, and better insight into the organization of the genome offered the opportunity to design approaches for the unbiased interrogation of the entire genome.¹³⁸ Sufficiently large case-control GWAS thus have the potential of identifying disease-associated mutations, even if the contribution of that polymorphism is minor. It should be noted here that GWAS often cannot unequivocally identify gene associations (at best “candidate genes”), as risk loci contain multiple genes, and, therefore, follow-up fine mapping or functional studies are generally required. Although GWAS have rapidly increased the list of known risk loci for many autoimmune disorders, some have questioned whether this knowledge substantially aids our understanding of their etiology and offers clues for rational intervention. It could be argued that all associations are often eclipsed by the major contributing factor, HLA polymorphisms, and essentially pinpoint genes with key roles in the immune system that could be expected to be involved to some degree. In T1D, for instance, minor associations with important genes encoding molecules involved in immunity such as CTLA-4 and IL-2 were discovered. From a therapeutic perspective, such findings may be hard to translate as, based on the relatively small association rates, these genes and their products may not necessarily represent pivotal targets in a diverse human patient population. In summary, it can be concluded that most likely all major susceptibility genes associated with the most common autoimmune disorders as discussed in the following text are currently known. Future years will tell whether this comprehensive genetic dataset translates to a better understanding of the origins of autoimmunity and if this will open new avenues toward better treatements.

Viral Infections. For many years, viral infections have been discussed as potential candidates to trigger autoimmunity in susceptible individuals because of their capacity to directly
infect target tissues and induce strong inflammatory responses and immune activation. While the association between viral infections and organ-specific autoimmune disorders is an intriguing possibility, it has been exceedingly difficult to demonstrate a causative role for specific viruses in human autoimmune diseases. Among the many obstacles are the fact that 1) all individuals undergo a multitude of viral infections during their lifetime; 2) one has to assume that viruses are frequently cleared at the time of diagnosis and viral footprints can be difficult to find in individuals affected by an autoimmune disease (“hit and run” event); 3) the precise viral strain, infection kinetics, and number of T cells and type of effector functions induced may play an instrumental role in determining its effect in an individual genetically at risk, necessitating a very detailed immunologic profiling¹³⁹; 4) due to MHC polymorphism, there is a significant variation in specificity of the antiviral response; 5) viral infections might not, per se, trigger autoimmunity but affect an ongoing autoimmune process in a detrimental way; and 6) there now is increasing evidence that viral and other¹⁴⁰ infections may rather prevent than enhance ongoing autoimmune responses, either by apoptosing aggressive T cells or by augmenting the T_reg pool.^141,142 Thus, a successful approach should be to first explore the underlying mechanisms in virally induced autoimmunity and then apply the precise insight and paradigms developed specifically to the human situation.

The most important mechanisms to be considered are listed here and are summarized in Figure 44.3.

The experimental evidence for all of these three scenarios obtained in different mouse models in vivo is well documented. However, none of these have been proven for any human autoimmune disorder due to the large size of human trials and the invasive nature of in vivo diagnostics required at the present stage. Thus, in the near future, we will continue to depend on animal models until noninvasive human in vivo diagnostic strategies have advanced and allow for imaging of trafficking of antigen-specific lymphocytes and high-resolution definition of immune process present in a specific organ. A final remark should be made for the existence of a negative association between viral infections and autoimmune disease found in several experimental models.¹⁴⁹ These observations are in support of epidemiologic findings that the incidence of many autoimmune disorders is decreased in equatorial countries, where the presence of certain infectious diseases is significantly increased. However, no firm associations have been established to date.

Other Environmental Causes. Similar to viral infections, other inflammatory stimuli may trigger or enhance autoimmunity. The gut deserves particular attention in this respect. At this site, each individual harbors thousands of different bacterial strains, and viral infections are common. Furthermore, the mucosal lining is permeable for nutrients and constitutes a very large interactive surface with the environment. Again, the complete absence of all bacteria results in severe immune dysfunction and possibly autoimmunity. However, it would be incorrect to conclude that infections are therefore always protective. Indeed, the commensal flora appears crucial in maintaining proper immune activation and function, but certain pathogens could definitely elicit strong gut immune responses that lead to autoimmune disease.^150,151

Cytokines and chemokines are essential regulators of cellular and humoral immune responses and lymphocyte trafficking.^152,153,154 They play a central role in orchestrating autoimmune processes and constitute a multitude of positive as well as negative feedback loops.¹⁵⁵ It is well established that certain cytokines can negatively or positively influence the production of other cytokines (ie, the Th1/Th2/Th17 paradigm) and thus determine the balance between proinflammatory and anti-inflammatory factors in the local environment. Furthermore, autocrine production feedback can augment or shut down production of a given cytokine by one cell. Cytokine networks operate with a fair amount of redundancy, and many cytokines and chemokines share common receptors. They are the most likely mediators of “bystander activation” and “suppression” processes and also offer an effective and versatile therapeutic target via the temporally restricted use of cytokine- or chemokine-blocking antibodies. Their precise function can vary quite dramatically in respect to the autoimmune disease under investigation and will be discussed in the section on Individual Autoimmune Disorders. Further, their level and timing of expression during an ongoing disease process will determine whether they have a positive or negative effect (or any at all; see the following discussion).

Many autoimmune disorders have traditionally been classified as “Th1-” or “Th2”-driven disorders based on, among other factors, the pathogenic role of cytokines such as IFNγ and IL-4.¹⁰⁷ Recent years have challenged this perspective with of the description the Th17 helper subset, characterized by the production of IL-17.^156,157,158 This class of T cells is considered highly proinflammatory, and its discovery was found to be of importance in several autoimmune disorders in both animal models and humans.¹⁵⁹ One example of an autoimmune disease where Th17 cells have taken center stage is RA. The disease’s principal animal model was long referred to as a Th1 driven, but the field struggled with the observation that mice treated with neutralizing antibodies to IFNγ- and IFNγ receptor-deficient mice developed more severe arthritis.^160,161 Dominance of Th17-directed immunity was put forward when IL-17 knockout mice were found
to develop less arthritis,¹⁶² and treatment with neutralizing antibodies to IL-17 or soluble IL-17 receptor alleviated joint inflammation.¹⁶³ In humans, it was found that IL-17 is increased in RA sera and synovial fluid and is present in the T-cell rich areas of the synovium.¹⁶⁴ Furthermore, IL-17 has profound inflammatory effects on a range of cell types within the joint. Several clinical trials are underway to either block IL-17 or the cytokines that govern the generation of Th17 cells such as IL-6.

It appears to be a general paradigm of great functional consequence that activation of the immune system is followed by a process that reverses the activation and reestablishes
homeostatic baseline levels of immunity.^88,90 In the absence of such regulatory mechanisms, immune responses will overshoot their goals and excessive immunopathology will occur. Thus, AICD is believed to play an important role in regulating autoimmunity. Apoptotic lymphocytes, for example, are easily detected in islet infiltrates in T1D,¹⁶⁵ and targeted induction of limited apoptosis may even prevent onset of autoimmune disease.¹⁶⁶ While increased apoptosis of aggressive lymphocytes that exceeds the “supply” of newly activated cells may directly limit an ongoing autoimmune process, limited apoptosis of target tissues may indirectly facilitate induction of protective regulatory responses. On the other hand, while apoptosis of target cells should at best be limited, decreased apoptosis of autoreactive aggressive lymphocytes will propagate autoimmunity. In RA, for instance, insufficient apoptosis of synovial macrophages, fibroblasts, and lymphocytes is one mechanism that might contribute to persistence of the disease and lead to synovial lining hyperplasia.¹⁶⁷ It is therefore important to consider precisely which cells undergo apoptosis in order to predict the possible outcome.

Thus, an ongoing autoimmune process can be viewed as a rather fine-tuned and fragile equilibrium of aggressive and regulatory components, and the precise activation kinetics and survival times of all lymphocyte types implicated in the process will determine the outcome. We are, at present, unable to delineate the precise in vivo cellular kinetics, and a more thorough understanding will require improved noninvasive diagnostic techniques.

One of the most important emerging areas for an improved understanding of the pathogenesis of autoimmunity is concerned with the kinetics of immune responses. The pathophysiologic or therapeutic effect of a given lymphocyte population depends not only on specificity, activation state, and effector functions but is also a function of the timing during which phase of an ongoing disease process it is present. Indeed, inflammatory cytokines such as IFNγ or TNF-α exhibit opposing effects in T1D depending on the precise time point of generation.¹⁷⁰ Early expression enhances islet destruction and disease development, whereas late expression ameliorates disease by inducing apoptosis of autoaggressive cells. These kinetic issues constitute a major obstacle for successful immune intervention because they preclude the use of specific blocking agents or administration of cytokines without precise knowledge of their kinetically differential role in the disease process. Figure 44.4 illustrates these kinetic considerations in relation to target cell destruction. A better understanding of the underlying “autoimmune kinetics” is essential and treatments will likely have to be individualized, in particular for antigen-specific immune-based interventions.

Treatment of autoimmune disorders is not that different conceptually from cancer therapy. A fine balance must be found between efficacy of the intervention and acceptable undesired effects. The main goal of autoimmune disorder therapy is suppression of the pathologic autoimmune response. Therapeutic options range in principle from continuous immunosuppression of the entire immune system to specific, targeted, temporally limited, and local immunosuppression. Systemic immunomodulation or anti-inflammatory therapy will affect the entire immune system and may compromise the immune status of the individual. One of the more succesful examples for effective systemic treatments is the blockade of TNF-α to ameliorate RA.^171,172 This new class of drugs is especially valuable because TNF blockade effects inflammatory pathways distinct from the targets of conventional anti-inflammatory therapy with corticosteroids or nonsteroidal anti-inflammatory drugs. TNF inhibitors can thus be used in patients that are refractory to the latter treatments with the added advantage of considerably improved outcome in terms of structural bone erosions. The remarkable therapeutic success of these drugs has since been replicated in other autoimmune conditions such as Crohn disease, ulcerative colitis, ankylosing spondylitis, and psoriasis. Whereas these diseases target different organs, common therapeutic improvements upon TNF blockade thus indicate shared mechanisms in terms of cytokine signaling. While undesired effects are relatively low, SLE-like symptoms have been observed in a few patients, as has the enhanced susceptibility to tuberculosis. Another important disadvantage, in addition to the high cost of these biologics, is that not all patients respond to TNF blockade. This has spurred a search for more potent blockers of inflammation, with the most promising developments stemming from the design of Janus kinase inhibitors for treatment of RA.¹⁷³ This class of small molecular drugs that interrupt signaling downstream of multiple rather than individual cytokines has recently cleared phase III trials in RA and is expected to revolutionarize the therapeutic landscape. While indications exist that this approach is at least safe in the short-term and that it outperforms anti-TNF biologics, long-term safety profiles obviously remain to be determined.

Autoantigen-specific immune interventions, in contrast, bear the promise of lower systemic side effects, as they can be targeted to antigens that are exclusively expressed in the diseased organ.^174,175 However, the efficacy might be lower and suitable target antigens have to be chosen carefully, because enhancement of autoimmunity is an important concern. The goal is either deletion of aggressive autoreactive T cells or induction of regulatory cells.¹⁷⁶ To achieve the latter, response modifiers are probably required at the time
of immunization in order to skew the resulting immune response to exert regulatory effector functions. Deletion of autoaggressive lymphocytes or anergy induction is even more risky, as only suboptimal immunization (ie, in the absence of costimulators) will result in this outcome. To control this in vivo is rather difficult. Ultimately, antigen-specific therapy will likely have to be individualized due to MHC polymorphism and distinct T-cell repertoires between individual patients and should be combined with other systemically acting agents, for example, antibodies against CD3 or CD4 or costimulation blockade. Indeed, there is intriguing evidence along these lines in animal models: Combination of a non-Fc-binding anti-CD3 antibody with mucosal immunization of insulin or insulin-derived peptides exhibited clear synergy and enhanced efficacy in reversing recent-onset diabetes in the NOD.¹⁷⁷ In this case, induction of insulin specific adaptive T_regs and their effector function were enhanced and protection by these T_regs was transferable and highly effective in that it reversed recent-onset T1D.

For anti-inflammatory interventions, the factor to be targeted should be as disease specific as possible. Therefore, blockade of TNF works well for RA but not for diabetes or MS.¹⁷⁸ Experimental evidence supports this observation because TNF is a crucial mediator found to be elevated in affected joints,¹⁷² but has clearly positive, as well as negative effector functions in murine models of MS and diabetes.^179,180 Targeting ubiquitously present chemokines or cytokines will likely not bear much success because of the resultant generalized immune modulation or suppression.

For antigen-specific interventions, antigens that are already targeted by regulatory autoreactivity are likely to constitute good targets to augment such a preexisting response.¹⁸¹ In contrast, these antigens should not be selected to delete autoreactive cells, which results in a loss of regulation. For anergy or apoptosis induction, antigens targeted by a primarily aggressive response will be better suited. The fact that autoantigenic and epitope spreading occurs during each ongoing autoreactive process makes such interventions difficult to design, and individualization will likely be necessary.

One of the factors that pose a challenge to understanding the pathogenesis of distinct autoimmune diseases may also hold a clue to developing effective and specific treatment strategies. Each target cell, tissue, or organ exhibits specific features that distinguish it from other sites of the body. These site-specific features of autoimmunity will
likely offer unique target sites for interventions with lower systemic side effects. However, one concern is that reestablishing proper immune homeostasis and regulation in one organ may still affect homeostasis systemically or at another site. Therefore, thorough preclinical evaluation and careful monitoring of undesired effects is urgently needed. Ideally, treatments have to be administered before complete organ destruction has occurred in patients identified by genetic or other screening to be at risk to develop full-blown clinical disease. During the preclinical state, frequently regulatory autoreactive responses are still strong and their augmentation can result in protection (Fig. 44.5). During advanced stages of autoimmunity, T_reg function or susceptibility of effector T cells to regulation might decrease.¹⁸² The goal to reestablish homeostasis and proper regulation after an initial insult that caused organ-specific inflammation appears a natural countermeasure to which the specific immune system may be successfully harnessed. An intriguing example is the emerging concept of organ repair that can potentially be enhanced by drug therapy. In T1D, regeneration and replication of β cells can possibly occur after giving exenatide, a GLP-1 agonist, which is an established intervention in type 2 diabetes and under investigation in combination with other drugs such as anti-CD3 in T1D.

Introduction and Disease Description. The year 1956 was a seminal year for the field of human autoimmunity given the discoveries of Hashimoto thyroiditis as an autoimmune disease and of Graves disease as caused by an autoantibody. These discoveries have prompted some straightforward and relatively uncomplicated treatments.^183,184 In Graves disease, autoantibodies directed against the thyroid-stimulating hormone receptor on thyroid cells stimulate excessive production of thyroid hormone, which is normally controlled by feedback regulation. In contrast, in Hashimoto thyroiditis, autoantibodies to thyroid peroxidase and thyroglobulin are present over years and likely able to fix sublytic doses of complement to cells of the thyroid. The result is an inflammatory reaction, which is also associated with T cell-mediated cytotoxicity. Thyroid damage due to painful Hashimoto thyroiditis may be associated with the development of Graves disease, indicating that there is a tendency for spreading of the autoimmune reaction in humans. Treatment of thyroiditis is relatively straightforward with antithyroid drugs (methimazole) and radioactive iodine.

Autoimmune Features. Autoantigens targeted in thyroiditis are thyroid peroxidase, a cell surface protein (Hashimoto thyroiditis), and the thyroid-stimulating hormone receptor (Graves disease). Autoimmune responses to thyroglobulin are also seen in animal models. The B-cell epitopes to these autoantigens have been mapped relatively well; however, as it is the case for other human autoimmune disorders, T-cell responses, their tracking and specificity, as well as their eliciting antigens in humans have remained largely elusive.¹⁸³ An interesting genetic link has been established to the CTLA-4 region, which is supported by the finding that NOD mice that also exhibit a link to this gene may present with thyroiditis. The usual suspects (ie, APC dysfunction, autoaggressive lymphocytes, and links to viral infections) have been examined, but no conclusive etiologic or mechanistic evidence has been obtained to date. It is noteworthy that a role for T_reg cells has been established in Mason’s animal model for thyroiditis.¹⁸⁵

Genetic Features and Environmental Factors. Interestingly, numerous viruses have been implicated in the pathogenesis of different thyroid diseases, but firm evidence for a direct involvement of viruses or virus-induced immune responses leading to clinically manifest disease is scarce. Subacute thyroiditis is a clinical and pathologic form of thyroid involvement that appears after infection with viruses such as measles, influenza, adenovirus, Epstein-Barr virus, and Coxsackie virus.¹⁸⁶ Again, however, a causative role in vivo has not been shown for any single infectious agent.¹⁸⁷ Presence of viral material in the thyroid and elevated virus-specific antibody titers were found to correlate with subacute thyroiditis.
In other instances, direct virally induced thyroiditis has been documented epidemiologically with thyroid or parathyroid disease.

Retrovirus, in particular human immunodeficiency virus (HIV) infections have generated much interest. Although HIV infection and acquired immunodeficiency syndrome (AIDS) may affect multiple endocrine organ systems,^{188,189,190,191} thyroid dysfunction usually reflects weight loss, anorexia, and cachexia of advanced HIV disease, rather than a direct viral effect on the thyroid.^192,193 Thus, direct involvement of the thyroid by HIV or by opportunistic infections is uncommon and may include subclinical hypothyroidism and “euthyroid sick syndrome.” The clinical relevance is probably limited, as overt hyperthyroidism or hypothyroidism does not occur with greater frequency in patients with HIV and AIDS as compared to patients with other nonthyroidal illnesses (HIV and the thyroid gland have been reviewed in Heufelder and Hofbauer¹⁹²). Involvement of the parathyroid in patients with AIDS could be shown by reduced basal and maximal parathyroid hormone levels, but the mechanisms underlying these findings have not yet been elucidated.¹⁹⁴ To account for a possible role of HIV in thyroid autoimmunity, a 66% homology between the HIV-1 nef protein and the human thyroid-stimulating hormone receptor has been noted. However, reactivity of sera of patients with Graves disease against a nef peptide showed no significant differences as compared to normal controls.¹⁹⁵ This does not rule out the presence conformationally shared T- or B-cell epitopes with HIV proteins. In analyzing mechanisms of molecular mimicry, studies of potential antigenic surfaces have emerged as an important supplement to analysis of sequence similarity.¹⁹⁶

In addition, antibodies from a patient with Graves disease showed reactivity to the gag proteins of another retrovirus, human foamy virus.¹⁹⁶ The association of human foamy virus with Graves disease or subacute thyroiditis is controversial. Whereas one study demonstrated human foamy virus-related sequences in the deoxyribonucleic acid (DNA) of peripheral blood in two-thirds of patients with Graves disease but none in normal controls,¹⁹⁷ another study could not confirm these findings.¹⁹⁸

In other studies, an association between human T-lymphotropic virus I and II and the occurrence of autoimmune thyroiditis or Graves disease was reported.^{198,199,200,201} Further, as in the case of HIV, hepatitis C virus was shown to lead to a wide variety of autoimmune disorders including involvement of the thyroid gland.^{202,203,204,205,206} Moreover, treatment of chronic hepatitis B and C with IFNα leads to induction or enhancement of autoimmune disease.^207,208,209 For congenital rubella infection, which has been associated with diabetes mellitus, Addison disease, growth hormone deficiencies, as well as thyroid disorders,²¹⁰ it is not clear whether thyroid involvement is the result of a direct viral effect or a more generalized dysfunction of the immune system.¹⁸⁷

Animal Models. Animal models in mice and chicken have been used to study virus-induced thyroiditis.¹⁸⁷ Mice persistently infected with LCMV showed reductions of thyroglobulin messenger RNA and circulating thyroid hormones in the absence of thyroid cell destruction.²¹¹ Thyroiditis characterized by focal destruction of the follicular structures, inflammatory infiltration, and generation of antibodies against thyroglobulin and thyroid peroxidase was observed in a reovirus type 1 mouse model of thyroiditis.²¹² The reovirus gene responsible for autoantibody induction was identified and the encoded polypeptide shown to bind to tissue-specific surface receptors.²¹³ Spontaneous lymphocytic infiltration of the thyroid is observed in the obese strain of chickens. Such chickens express an endogenous retrovirus, avian leukosis virus (ev22), not found in healthy normal inbred strains.²¹⁴ Although ev22 appears to be a genetic marker rather than cause for thyroiditis, infection of normal chicken embryos with avian leukosis virus can cause hypothyroidism.²¹⁵ Moreover, aberrant MHC class II expression is demonstrated in obese strain chickens and elevated levels of 2.5-oligoadenylate synthetase as well as 2,5 oligoadenylate-polymer levels in the cytosol of thyroid epithelial cells occur, suggesting viral involvement.²¹⁶ Again, not all cross-reactivities with self-ligands need to increase autoimmunity,²¹⁷ and regulatory cells also play a major role in modulating autoimmune thyroiditis.¹⁸⁵

Introduction and Disease Description. While the distinct symptoms of diabetes mellitus have been known since antiquity, the underlying pathophysiologic processes were only identified in the late 19th and early 20th centuries. The proof of the involvement of the pancreas in diabetes etiology was conducted by von Mering and Minkowski,²¹⁸ who demonstrated in 1890 that extirpation of the canine pancreas results in the classic symptomatology of hyperglycemia, abnormal hunger, increased thirst, polyuria, and glycosuria. Subsequently, inflammatory changes in the endocrine pancreas (ie, the islets of Langerhans) were correlated with diabetes by Schmidt²¹⁹ in 1902; two decades later, Banting and Best²²⁰ identified insulin as a pancreatic hormone, thereby providing the basis for insulin substitution therapy, which remains to this day the cornerstone for T1D management. In a classic 1965 paper, Gepts²²¹ noted the histopathologic similarity between thyroiditis and insulitis and suggested an immune basis for the disease. By 1974, the concept of T1D as an autoimmune syndrome was firmly established by the discovery of islet cell antibodies and an association between T1D and certain HLA genes.^222,223,224

Autoimmune Features. Today, a quarter century later, the possible autoimmune origin of T1D is understood in much greater detail. However, the lymphocytic infiltration of the islets of Langerhans (Fig. 44.6) and the presence of antibodies specific for β-cell antigens associated with the progressive destruction of insulin-producing β cells^225,226 still constitute the cardinal evidence for an autoimmune etiology. While there is a reasonably strong genetic linkage to certain HLA molecules, the disease has to be considered polygenic in nature,¹³⁰ and a significant discordance of disease among monozygotic twins suggests that environmental factors contribute to trigger and/or exacerbate the disease. Furthermore, it remains unclear which islet antigens are the
primary targets. The earliest islet cell-specifc antibodies in human individuals at risk are directed to insulin,^227,228 and evidence from relevant animal models and human studies points toward insulin as a primary antigenic target,²²⁹ islet-specific glucose-6-phosphatase catalytic subunit related protein,^230,231 and possibly glutamic acid decarboxylase (GAD),²³² but definitive proof for a pathogenic role has only been ascertained in the NOD mouse model and not yet in humans.²³³ The cellular infiltrates found in the islets contain both CD4+ and CD8+ T lymphocytes and their irreducible role in β-cell destruction has been documented in several animal models.^234,235,236 CD8+ T cells can exert direct cytotoxic effects toward MHC class I-expressing β cells, while CD4+ T cells secreting inflammatory cytokines and can provide help to CD8+ T cells as well as B cells. Ultimately, it is important to consider that β cells constitute about 60% of the islets, which in turn contribute only -2% to the pancreas mass and demonstrate, unlike many other tissues targeted in autoimmune disorders, an exquisite sensitivity to cytokines such as IL-1, TNF, and IFNs that will result in their apoptotic cell death after prolonged exposure.⁹⁴

The detection of islet antigen-specific antibodies remains an essential tool in identifying prediabetic subjects and monitoring the progression of subclinical and clinical disease. Procedures for autoantibody determination have been substantially refined and standardized worldwide. Emerging data from clinical studies support the notion that with progression of the prediabetic phase, generation of islet antibodies also is increased.^237,238,239 Usually, antibodies to insulin become discernible first, then to GAD, thereafter to insulinoma antigen 2. Individuals with islet antibodies to three or more distinct antigens have a greater than 90% risk of developing T1D.¹³⁶ Thus, islet antibodies are an excellent marker for disease risk. However, they appear not to play a role pathogenetically, as transfer of antibodies from mothers to children does not increase the risk for T1D, and B cells are not needed for human diabetes.^240,241,242 In this crucial respect, the NOD mouse appears to provide a paradigm that might not be applicable to human diabetes, as maternal antibodies are an essential factor for diabetes development in NOD offspring.²⁴³

Measurements of human T-cell responses to islet antigens are not yet standardized and can vary considerably between different laboratories. One reason for this may be the source of T cells that are generally subjected to specificity analyses: blood-borne CD4+ or CD8+ T cells may not reflect the specificity distribution and frequencies of isletspecific T cells found in the target organ (ie, pancreas and its draining lymph nodes). Even the study of spleen-derived islet-specific T cells readily obtained from NOD mice and analyzed in standardized proliferation assays has shown variations between different NOD colonies.²⁴⁴ Therefore, measurement of multiple effector functions (ie, cytokine production, etc.) in highly standardized assays will likely be required to assess T-cell autoreactivity on a routine basis.²⁴⁵ A recent study shows that while techniques such as enzyme-linked immunosorbent spot can discriminate between patients and controls, difficulties in detection of low-frequency antigen-specific T cells probably leads to the limited reproducibility of such measurements.²⁴⁶ Another complicating fact is that T-cell responses between individuals are expected to vary and depend on the HLA haplotype and individual trigger(s) that precipitate T1D. An emerging technology that may improve specificity and sensitivity is through the use of combinatorial HLA multimers loaded with multiple disease-associated epitopes.²⁴⁷ A remarkable paper²⁴⁸ has shown that tracking of immune responses to naturally processed peptide epitopes can discern between healthy individuals and those afflicted with T1D: Whereas the number of overall T cells reacting with the naturally
processed peptide epitopes was similar, individuals with T1D produced more IFNγ relative to IL-10, whereas the ratio was the opposite in healthy individuals.

Genetic Features and Environmental Factors. Recent data show that concordance rates among monozygotic twins are higher than previously thought.²⁴⁹ Still, the remaining discordance and particularly the major temporal differences in clinical onset indicate that environmental factors have to act in concert with diabetes susceptibility genes to orchestrate the autoimmune destruction of β cells. The initial hope that only a few genes would contribute to disease pathogenesis and that genetic links would help to directly understand the mechanistic aspects of T1D pathogenesis has been progressively eroded. Instead, a complex network of susceptibility and resistance genes in both humans and animals (eg, the NOD mouse) has slowly taken shape. The recent publication of three GWAS has extended the list of gene associations in T1D. So far,¹⁶ robust T1D-associated loci were discovered, some of which encompass important immunity genes (eg, IL-2, IL-2Rα [CD25], CTLA-4) or β cell-specific products (insulin), while other candidate genes do not have an apparent link to immune functions and/or β cells. The implications of this enhanced genetic characterization have so far not led to the identification of monogenic pathways that directly lead to disease. Rather, consensus is growing that disease susceptibility is the consequence of a global problem of immune regulation, with many genes involved in a variety of combinations in individual patients.²⁵⁰

Many of these genes exhibit direct parallels to NOD diabetes susceptibility genes.²⁵¹

The association between particular HLA/MHC class II haplotypes and the occurrence of human diabetes has been of particular interest. The DRB1/04-DQA1/0301/B1 and DRB1/03-DQA1/0501/0201 strongly predispose to T1D, and more than 80% of patients carry either one or both alleles.²⁵² In contrast, other MHC class II haplotypes can protect from disease, as evidenced by the sixfold reduced risk to T1D in DRB1/15-DQA1/B1/0602-bearing individuals. Considerable evidence indicates that other genes in the MHC region likely contribute significantly to T1D risk.¹²⁸

An intriguing mechanistic hypothesis was put forth by McDevitt’s observation that predisposing HLA class II alleles appear to express small neutral amino acids at position 57 of the DQ allele of Caucasoid populations, whereas an aspartic acid is found in resistant alleles at the same position. Because position 57 is part of the peptide-anchoring pocket, amino acid substitutions in this area will affect peptide binding. Indeed, the susceptibility alleles prefer different peptides, but the contribution to T1D development is not yet clear and a mechanistic link has to be established. Both central and peripheral tolerance mechanisms have been implicated but no direct proof has been obtained. It is important to realize that the human susceptibility MHC class II alleles share amino acids at position 57 with the I-A^g7 alleles expressed in the NOD mouse and are required for NOD T1D predisposition. However, as previously mentioned, polymorphisms in the MHC class II coding region alone cannot explain diabetes pathogenesis. The amount of complexity involved in the immunogenetics of T1D has been well described by Serreze²⁵³:

Thus, diabetes susceptibility and resistance genes contribute to disease in a polygenetic/multifactorial fashion that appears to gain in complexity as it is being unraveled. The link to environmental factors will be defined to shape gene expression and disease development. Major contributors in this respect appear to be the gut and viral infections.

With more than 400 m² of mucosal epithelium, the gut constitutes the largest interactive surface area of the human body connecting us with the environment and its pathogens.¹⁵⁰ Therefore, exposure to antigens or pathogens through the gut, mediated by the largest outpost of the immune system, the gut-associated lymphoid tissue, will strongly affect specific and general immune functions. It is intriguing that immune tolerance to the numerous foreign protein antigens found in food, as well as bacterial antigens derived from the commensal flora, is generally well maintained.²⁵⁴ This may be attributable to the high levels of immunoglobulin (Ig)A and TGF-β in the gut and to the phenomenon of “oral tolerance.”²⁵⁵ Oral tolerance has been observed in animal models and humans and is characterized by tolerance induction to protein antigens present in the gut. It occurs via two principal mechanisms: Low amounts of antigens will induce a nonaggressive immunoregulatory response while high amounts of antigen can lead to lymphocyte anergy or deletion,^256,257,258 which is likely achieved via APC modulation. In addition, the profound immune dysregulation found in the absence of a bacterial flora in both animals and humans points to an important physiologic role that foreign antigens play in immune homeostasis in the gut.²⁵⁹ Furthermore, NOD mice only exhibit high levels of autoimmunity when kept in a clean, specific pathogen-free environment; they do not develop T1D when housed under “dirty” conditions.²⁶⁰ A seminal paper by Wen et al.²⁶¹ highlighted the pivotal role of the mucosal immune compartment by showing that host recognition of the digestive flora is essential in preventing T1D through engagement of a myeloid differentiation primary response gene 88-independent signaling pathway. Thus, changes in the intestinal microbiome, integrity of the mucosal barrier, infections of the gut, or certain dietary components may play a role in T1D pathogenesis.

Several reports and studies have attempted to establish a link between the introduction of cow’s milk and development of T1D in young infants. This link was not observed in the German, Australian, and American baby diabetes studies but was in a Finnish epidemiologic study.^262,263 The Finnish study differed from most of the others by an
extended observation time involving infant as well as childhood consumption of cow’s milk. Therefore, a dietary link between milk feeding and T1D can be considered unlikely but not excluded after long-term exposure to cow albumin or other milk proteins. Similarly, wheat-derived gluten and milk-derived insulin have been implicated as a cause for childhood diabetes. The evidence, however, is not convincing at this point, and no firm links have been established.

Some intriguing observations have been published more recently supporting the concept of a viral etiology for T1D. The mechanistic links between viral infections and autoimmunity can be manifold and have been discussed in detail in the introductory section of this chapter. A significant association between rotavirus infection in young infants and the first occurrence of islet autoantibodies was established by Harrison’s group in Australia²⁶⁴ but not in Finnish populations.²⁶⁵ Rotavirus is a double-stranded RNA virus, infects the intestinal mucosa, and is a common cause for seasonal childhood diarrhea. It can polyclonally activate T and B lymphocytes, and might possibly harbor antigens that could immunologically mimic islet cell-derived self-proteins. However, it is not clear whether it infects the pancreas or islets directly. The most convincing case can be made for enteroviruses, and there is now robust evidence for a significant association with T1D.²⁶⁶ Coxsackie B4 virus has been isolated from islets of a child with acute-onset T1D,^267,268,269 and Coxsackie B3 and 4 strains commonly infect the gut, pancreas, and heart.²⁷⁰ They lead to profound pancreatitis if they replicate at high enough titers and might harbor a mimicry antigen (P2C protein)²²⁶ cross-reactive on the T-cell level with a human GAD epitope. However, this evidence could not be replicated by other laboratories and is still controversial. It has become apparent, however, that Coxsackie B3 viral strains can effectively trigger²⁷¹ or prevent¹⁴² T1D in the NOD mouse, depending on the timing of infection. Thus, diabetogenicity of a viral infection may critically depend on the preexistence of insulitis and may either act as an initiating factor or by aggravating ongoing inflammatory processes. Clinical studies have recently provided indirect support for both of these scenarios.^272,273 The finding that infections under certain conditions confer protection might support the “hygiene hypothesis,” which suggests that infections protect from rather than enhance autoimmunity. It is still uncertain whether enteroviruses routinely infect or persist in pancreatic islets, although recent reports indicate that viral particles are found more frequently in islets of patients with T1D around the time of disease onset.²⁷⁴ Similar to Coxsackie, other enteroviruses such as polio or echoviruses have been detected in the pancreas and might therefore at least have enhancing effects on ongoing islet destruction in prediabetic individuals at risk.²⁶⁷ The establishment of a firm association between viral infections and T1D is difficult because the underlying mechanistic links established in several animal models allow for the virus to be cleared before autoimmunity develops (ie, in the rat insulin promoter [RIP]-LCMV model); viruses need not necessarily directly induce islet-reactive T-cell responses but can act as bystander activators and, in many cases, viral infections have been found capable of preventing autoimmunity.

Exciting new data suggest that the genetic constitution may (co-)determine whether viral infections will provoke islet autoimmunity in certain individuals. It was found that certain rare polymorphisms in the IFIH1 gene are associated with protection against T1D,²⁷⁵ while others confer disease risk.²⁷⁶ This gene encodes a helicase enzyme, IFIH1 (also known as MDA5), which triggers the secretion of IFNs in response to viral infection. Most, if not all, viral infections trigger the production of IFN by the host immune system. Being an IFN-response gene, IFIH1 allows the infected cell to sense the RNA genome of enteroviruses and increase IFN production. IFNs then primarily limit viral replication to prevent damage to the infected cell. They also, however, increase the visibility of the infected cell to the immune system by enhancing the expression of MHC molecules. A scenario could be envisioned where mutations leading to decreased IFIH1 activity are associated with milder responses against diabetogenic viruses and thus disease protection. Conversely, mutations leading to IFIH1 hyperactivity could induce inappropriate, exaggerated IFN responses and localized immunity against the infected β cells. Altogether, such a mechanism would explain why viruses that have presumed diabetogenic effects do not necessarily trigger diabetes in all infected individuals.²⁷⁷

Animal Models. Because the pancreas and its draining lymphoid organs are notoriously difficult to access, many important insights about diabetes immunology have been gained from suitable animal models that continue to refine our understanding of the pathogenesis and the development of potential prophylactic and therapeutic strategies. There are a multiplicity of animal models for T1D. The most commonly employed models take advantage of natural mutations that give rise to spontaneous diabetes onset or antigen-specific induction of disease using transgenic technology. Other models make use of β-cell damage initiated by treatment with specific chemicals (eg, streptozodozin) or virus infection. Encephalomyocarditis virus is diabetogeneic in mice; the incidence of disease is dependent on both virus and mouse strains used.^{144,268,269,278,279,280,281,282,283} Similarly, Coxsackie virus, associated with diabetes development in humans, causes extensive pancreatic tissue damage and release of sequestered autoantigens that lead to rapid diabetes development in some mouse strains.²⁸⁴

Models of Spontaneous Diabetes Onset. There are several animal models of spontaneous T1D.²⁸⁵ The two most extensively used are the biobreeding rat, introduced in 1974 at the Bio Breeding Laboratories in Canada, and the NOD mouse strain established 1974 in Osaka, Japan.²⁸⁵ Because the biobreeding rat is associated with leukopenia and other abnormalities,²⁸⁵ the NOD mouse has been the model of choice due to its genetic linkages that are reminiscent of human T1D.²⁸⁵ In both models, adoptive transfer of T cells can induce disease.²⁸⁶ Interestingly, viral infections, first shown with LCMV, can prevent insulindependent diabetes mellitus in both biobreeding rats and NOD mice.^287,288 This occurs in the absence of a general immune suppression. While the mechanism involved is unclear, the generation of suppresser T cells has been suggested.^288,289