seminars in IMMUNOLOGY, Vol 10, 1998: pp 79]86

Protective and destructive effectsof microbialinfection in insulin-dependent diabetes mellitus

Bhagirath Singh, Sean Prange and Anthony M. Jevnikar

( )Insulin-dependent diabetes mellitus IDDM is a T-cell me-diated autoimmune disease, which results in the destructionof the islet b-cells. The major histocompatibility complex( )MHC encodes the major susceptibility gene in IDDM. Theconcordance rate for diabetes in identical twins is 30]50%and in inbred animal models of disease the incidence rate is20]80%. These results emphasize a role for environmentalfactors in the disease process. It has long been suggested thatIDDM in humans may be caused by viral infections. Whileconsiderable progress has been made in defining the geneticsof IDDM, our understanding of the role of environmentalfactors, which might provide a more direct approach to

( )therapy is considerably lacking. We suggest that 1 thedensity and affinity of epitopes derived from microbial anti-

( )gens that bind to MHC molecules; 2 their cross-reactivity( )with b-cell antigens; and 3 the nature of immunoregula-

tory cytokines induced by the microbial infections are theprimary factors in the induction of either effector or protec-tive T cells in IDDM.

Key words: Immunopathogenesis r microbial infection rBCG immunotherapy r immunoregulation r tolerance rIDDM r type I diabetes r NOD mice

Q1998 Academic Press Ltd

Ž .INSULIN-DEPENDENT DIABETES MELLITUS IDDM is per-haps the prototypic example of an autoimmune dis-ease resulting from the interaction of susceptibilitygenes and the environment. While the risk of devel-oping disease increases in families with autoimmunediabetes, the concordance rate in identical twins is30]50%, clearly emphasizing a potential role for en-

From the John P. Robarts Research Institute and the Depart-ments of Microbiology, Immunology and Medicine, University ofWestern Ontario, London, Ontario N6A 5C1, Canada

Q1998 Academic Press Ltd1044-5323r98r010079q08 $25.00r0r si970107

vironmental factors. However, while considerable hy-potheses have been forwarded to involve the causativerole of infectious agents in disease, this concept is insharp contrast to the actual increased incidence ofdisease in pathogen free NOD mouse colonies. Theseseemingly disparate observations might be reconciledby a more broad concept in which there are bothcausative and protective responses to microbial agentsandror their products in diabetes.1 Unfortunately,while considerable progress has been made in de-fining the genetics of insulin-dependent diabetesmellitus, little work has been done to advance ourunderstanding of the role of environmental factors,which might provide a more direct approach to thetherapeutic strategies.

Role of the environmental factors in IDDM

The debate over the role of environmental agents inIDDM was initially based on observational studies ofIDDM following various infections. The hypothesisthat viral infection may result in IDDM was firstintroduced more than a century ago by Stang whoreported a case after chicken pox.2 Of particularinterest are rubella, coxsackie and cytomegalovirusinfections. At least in one case, virus was isolatedfrom pancreas at the time of autopsy and in others,anti-viral antibodies were detected in the sera at thetime of diagnosis.3 However, there is no conclusiveevidence that IDDM in humans results from a viralinfection.4

The development of diabetes in humans and NODmouse is polygenic with susceptibility related to bothMHC and non-MHC loci. The involvement of variousdiseases’ susceptibility genes in IDDM has been re-cently reviewed both in humans and rodent models.5,6

However, these studies have not addressed the con-tribution of various environmental factors in the dis-

79

B. Singh et al

ease process. These include bacterial and viral infec-tions, diet and seasonal variation.1,7,8 Interestinglythere are striking differences in the incidence ofType 1 diabetes between different countries withScandinavian countries having the highest relativenumbers of 30 per 100,000 each year 2 suggestingthat there may be an inverse association between thedevelopment of diabetes and socioeconomic develop-ment. A recent study suggests that dietary compo-nents may also contribute to the disease process.9

One of the important links between these observa-tions and IDDM is the immune system. Given themillions of years of co-evolution of our immunesystem with the microbial world, it is not surprisingthat there could be a direct influence of microbialagents on autoimmune diseases. Many of the envi-ronmental factors can influence the immune systemand could control the induction and progression ofIDDM.10 Cross reactivity between microbial proteinsand the b-cell antigens,4 the immunoregulatoryproperties of viral proteins and the production ofcytokines or chemokines following microbial infec-tion could affect the induction of or protection fromthe disease.

Microbial infections have a capacity to regulatediabetes both positively and negatively. A survey ofNOD colonies throughout the world has revealedthat there is a wide range of diabetes frequency invarious colonies,11 probably due to environmentaleffects with a greater incidence of diabetes in thosecolonies tested to be specific pathogen free. Therange appears to be anywhere from 20 to 90% infemales at age 30 weeks, and 1]65% in males.6 Thereis a reciprocal relationship between the incidence ofdiabetes and the level of infection within the colony.11

Viral infections, molecular mimicry andsuperantigens in IDDM

Viral infections

It has been considered for some time that micro-organisms in their capacity to both stimulate theimmune system and express multiple antigens, maybe involved in the initiation andror progression ofautoimmune diseases. Structural homology of viralantigens to self-antigens might lead to cross-reactivedestructive immune responses in a process referredto as molecular mimicry.12 This has raised the obvi-ous question of whether there is a role for molecularmimicry in autoimmune diabetes?4

Research in the 1960s identified a possible linkbetween coxsackie virus and diabetes by finding aseasonal variation between neutralizing antibodies tovirus type B4 and IDDM. In a recent analysis of 128patients with new onset insulin-dependent diabetesmellitus, diabetic children were more likely than con-trol patients to be IgM positive for nine of 14 serotypesof coxsackie virus B1, B2, B6 and A9, ecoviruses 4, 6,9, 11, 30 and 34 and enterovirus 71.13 While theseand other data support an association between IDDMand enterovirus IgM positivity in some children2,13

and despite more than three decades of research inthis area, the question remains whether enterovirusesare central to the autoimmune process or whetherantibodies to viral proteins occur by immunologicalcross-reactivity but are unrelated to the developmentof diabetes. There are about 67 distinct entroviralserotypes of antibodies which further complicates theanalysis,14 and most studies have focused on relatively

Ž .acute infectious events - 6 months using IgMtitres.

There may be direct effects and indirect conse-quences of viral infections. In a simple mechanism ofcellular injury leading to diabetes, viruses might re-duce b-cell mass by a single severe or repeated infec-

Ž .tions Figure 1A . The involvement of viral pancreaticinfections would be consistent with data that demon-strate early insulitis is associated with CD8 and mono-cyte infiltrate, and that CD4 infiltration is a laterevent.4 Alternatively, in the concept of molecular

Žmimicry, infections with coxsackie rubella or cy-.tomegalovirus virus may lead to disease if viral pro-

teins share amino acid sequence homology with cer-tain b-cell autoantigens which then could lead tob-cell destruction via cytotoxic T cells or activateCD4q T cells that mediate insulitis.4,15,16 Obviouslythere has been an intensive search of databases ofprotein sequences to determine if there are homolo-gies between certain micro-organisms and humanself proteins.

There are a number of conceptual problems withthis hypothesis.4 Pancreatic tissue from recent onsetdiabetic children did not demonstrate viral genomesof candidate viruses such as CMV, mumps, rubellaand coxsackie virus by RT-PCR.17 In a well-describedanimal model of diabetes,18 inoculation of BB rats

Ž .with Kilham rat virus KRV appears to activate dia-betogenic T cells rather than directly injure islets.16

This possibility has not been unequivocally ruled outas mumps virus, coxsackie virus B3 and B4, and reovirus type 3 can infect human cells in vitro 15 and

80

Microbial infection and IDDM

Ž .Figure 1. Mechanisms of the activation of islet b cell reactive T cells in IDDM by microbial infection. A Direct damage toŽ . Ž .b cells left by viral infection resulting in the release of autoantigens and their subsequent presentation on APCs. B

Antigen processing of microbial antigens could generate epitopes that activate islet b cell reactive CD4 andror CD8 T cellsŽ . Ž .due to molecular mimicry. C Activation of islet b cell reactive T cells by microbial superantigens. D Competitive capture

of diabetogenic epitopes by IDDM associated DQrIAg7 and protective DRrIE molecules influences the development of theŽ . Ž . Ž .disease. E Presentation of diabetogenic epitopes with higher left or lower right densityraffinity for MHC molecules in

the activation of autoreactive T cells in IDDM.

there has been isolation of coxsackie B virus frompancreas of some children with acute onset IDDM.19

Molecular mimicry

There is evidence linking the b-cell autoantigenswith microbial epitopes in IDDM through molecularmimicry.4 Cross-reactive epitopes may be generatedby antigen processing which may be presented by thehost MHC molecules to the auto-reactive T cellsŽ .Figure 1B . IDDM is clearly a T-cell mediated dis-ease with considerable experimental data to supporta role for CD4 cells, CD8 cells and macrophages.10

The function of diabetogenic lymphocytes is influ-enced by the cytokine micro-environment, which inturn may be influenced or altered by environmentalfactors such as microbial infection.1,20 Self-reactivelymphocytes may escape the process of negative se-lection within the thymus but they are regulated byperipheral mechanisms that prevent autoimmune ex-

pression. In some individuals, T-cell responses thenare directed against b-cell antigens which in turnleads to eventual b-cell death and glucose intoler-ance. However, while a panel of auto-reactive pro-

Žteins have been identified reviewed by Zechel et21.al , insulin is the only one that can truly be con-

sidered b-cell specific.4 The appearance of cell me-diated andror antibody responses against self-pro-teins may herald loss of tolerance and eventual dis-ease. We have found that many of the epitopes de-rived from self proteins22 including MHC class IImolecules23,24 are highly immunogenic in vivo. Therecognition of MHC determinants and the role ofTCR]antigen]MHC interaction is complex, and evensingle amino acid alterations may sufficiently alter thepeptide binding motif within the MHC bindinggroove to either positively or negatively influence thefull expression of disease.25 Earliest detectable im-mune responses to self-proteins in the NOD mice,however, appear to be towards glutamic acid decar-

81

B. Singh et al

Ž .boxylase GAD , but with time involve multiple epi-topes within the protein and eventually with b-cellinjury, other islet cell proteins such as heat shockprotein 60, carboxypeptidase H and insulin.26 Thismay represent both intra-molecular spreading withantigenic spread from one epitope to others withinthe same protein and inter-molecular spreading toother proteins, both of which have been demon-strated in the NOD mouse model.6,26,27

There may be cross-reactivity between rubella viruscapsid protein and a 52 Kd islet antigen28 and asequence similarity exists between the 40 Kd cox-sackie virus P2]C protein and the autoantigen GAD65which is highly expressed within human pancreas.The sequence homology between human GAD65 andcoxsackie virus P2]C protein contains a T-cell epi-tope involved in GAD responses in humans withIDDM.29 GAD65 autoantibodies exist in patients withIDDM as well as in those at risk for the disease. T-cellresponses in newly diagnosed IDDM patients directedagainst amino acid regions 247]266 and 260]279obviously overlap with the 250]273 region of GADwhich contains the sequence homology to P2]C pro-tein of coxsackie B virus.4,14,16 Furthermore, immu-nization of NOD mice with coxsackie virus P2]Cprotein can induce T-cell responses in the mice thatcross-react with GAD or GAD peptides.30 The criticalinvolvement of MHC is emphasized by the ability todemonstrate these T-cell responses only in mice ex-pressing the I-A g7 haplotype found in NOD mice.30

However, although the P2]C sequence seems to bethe target of diabetes associated autoantibodies, antiGAD65 mAbs showed no cross-reactivity with nativeviral antigens of coxsackie B1]6, rubella or cy-tomegalovirus, emphasizing that the story remains farfrom clear. It is possible that detectable levels ofantibodies to relatively common enteroviruses in someindividuals may actually represent increased humoralimmunity linked to specific HLA genotypes and pre-vious studies have not adequately adjusted case-con-trol comparisons.14

In addition processing of viral proteins in vivo maynot actually yield immunogenic epitopes. The modeldoes not explain how the autoimmune disease ismaintained following loss of inductive antigen, in theabsence of repeated infections. It may be that there isintra-molecular determinant spread of antigen reac-tivity 31 which leads to a process no longer requiringthe initial antigen.4 Finally, one needs to considerhow in molecular mimicry, auto-reactive T cells aremaintained in an unresponsive state.

Superantigens

Is there a role for superantigens in the induction ofinsulin-dependent diabetes mellitus?15 Superanti-gens, unlike conventional antigens, function by bind-ing outside of the MHC binding groove on the lateralaspect of the MHC]TCR complex and directly inter-

Žact with the Vb portion of the TCR molecule Figure. 32,331C . By doing so, they have the capacity to broadly

influence T-cell reactivity in autoimmune diseases.While several systemic viral infections can injure isletsand cause mononuclear cell infiltrate, a direct linkbetween these viruses and cytopathic effect on b cellsis lacking.15 However, superantigens in the form ofbacterial products or retroviruses are widely repre-sented in our environment and may influence IDDMin the absence of direct cytopathic effects.

Retroviruses have been suggested but not beenwidely studied for a potential role in IDDM.15 How-ever, they may have the capacity to modify the selec-tion of the T-cell repertoire by their capacity tofunction as superantigens. For example Mls genescan stimulate a large portion of T cells in MLR fromMHC identical strains of mice.34 While superantigenbinding to MHC is not MHC restricted, there aredifferent binding affinities between HLA class II alle-les. While neonatal mice may be influenced by Mlsproducts of endogenous genes or maternally trans-ferred early in life, T-cell repertoires in adult micemay be altered by exposure to bacterial derived

Ž .products such as staphylococcal enterotoxin B SEB .There is recent evidence for endogenous viral super-antigen activity in humans.35,36,37 Pancreatic tissueand spleen from newly diagnosed Type 1 diabetics

Ž .showed an exceptionally high )30% expression ofthe TCR Vb 7 family compared to spleen or periph-

Ž . 36,37eral blood of healthy individuals 8]11% . Repeti-tive stimulation of peripheral blood lymphocytes fromnon-diabetic individuals with islet protein led to asimilar several-fold increase in the expression of Vb 7positive clones which together suggest that T-cellproliferation was driven by antigen present on theislet cells. Analysis of CDR3 in cloned T cells frompatients support the involvement of a super-antigen.15,36 Interestingly, high levels of reverse tran-scriptase in both islet and spleen of patients suggestthat a retro virus might be involved in the activationof infiltrating lymphocytes rather than the pancreaticb cells. So in contrast to molecular mimicry with viralor bacterial proteins, superantigens may be able tostimulate large proportions of the immune compe-

82

Microbial infection and IDDM

tent T cells leading to polyclonal activation. Despitethese results a definitive role of viral superantigens inIDDM still remains to be established. In NOD mice,in vivo treatment with the bacterial superantigen,

Ž .streptococcal entrotoxin B SEB prevents the devel-opment of IDDM.38 This treatment probably inducesregulatory cells which prevent induction of diabeto-genic T cells.

Immunoregulation and prevention of IDDM

Since IDDM is a T-cell mediated autoimmune dis-ease, these cells must recognize self-immunodomi-nant epitopes in association with MHC molecules.Therefore, MHC antigens must play a significant rolein the disease process. The disease pathogenesis is atwo step process involving the activation and homingof T cells to the pancreas in a slow chronic fashionfollowed by a rapid disease due to b- celldestruction.39,40 Major immunological insight in thedisease process has come from the work done in the

Ž .nonobese diabetic mouse NOD which is a seren-deptously discovered spontaneous model of autoim-mune diabetes and a widely accepted model for hu-man Type I diabetes.6 NOD mice develop insulitisearly by 4]6 weeks of age and overt diabetes occursin 50]90% of female mice by 30 weeks. The NOD

Ž d g7class II region of NOD mice is unique K , I-A ,nul l b. 6I-E , D . Due to a deletion and mutation in the

promoter region of the E a chain, NOD mice do notŽ g7 .express I-E and also have a rare I-A I-A molecule.

The I-A a chain is identical to H-2 d but the b chainhas a serine at residue 57 rather than aspartic acidfound in most mouse strains. The b chain is knownto contribute to the immune responses in NODmice.41 Similarly, resistance to type I diabetes inpatients is also associated with aspartic acid at posi-tion 57 of the DQb chain.42 The phenotype of acti-vated T cell is largely determined by the peptidesbound in the antigen binding groove of the MHCmolecules and the cytokines present10 at the time ofpriming. In the last few years, many investigators havecharacterized immunoregulatory auto-MHC class II

Žreactive T cells in NOD mice reviewed by Chaturvedi24 .et al . NOD mice lacking class I MHC molecules

Ž .b ]microglobulin-null are completely resistant to2IDDM suggesting a obligatory role of CD8 cells in thedisease process.6

The H-2 g7 NOD mice of which encodes for theIdd-1 gene has a moderately high penetrance for thegeneration of insulitis but unless homogygous has a

low penetrance for the development of spontaneousdiabetes.6 This view would be consistent with recentdata from our laboratory which demonstrates that

Žmice with only a single ‘dose’ of NOD class II class IInull .deficient NOD-I-A =NOD have significant insuli-

Ž .tis but rare overt diabetes unpublished observation .Introduction of non-I-A g7 MHC class II genes in theNOD background by breeding or transgenic technol-ogy 6,43 abrogates diabetes and insulitis. Therefore,the combination of the unique I-A g7 and the lack ofI-E expression are both important for the develop-ment of diabetes.6 The expression of non-I-A g7rDQMHC class II molecules such as I-E or DR couldprevent the presentation of diabetogenic epitopes by

Ž . 44,45determinanat capture Figure 1D .In the context of the MHC class II molecules,

microbial products are likely involved in the activa-tion of CD4 Th1 or Th2 cells andror CD8 CTL cellsin IDDM. Microbial products could have a bystandereffect on T cells by activation of monocytes and other

Ž .non-lymphoid cells basophils, eosinophils, etc.which can produce various regulatory cytokines. Pro-duction of IL-12 by mono-nuclear cells has beenshown to be involved in Th1 cell activation while IL-4induces the activation of Th2 cells.46.47 In NOD micethere is strong evidence that Th1 cells producingIFN-g and IL-2 are diabetogenic whereas the Th2cells which produce IL-4, IL-10, etc., are not. It hasalso been found that the administration of IFN-gaccelerates the disease while IL-4 is protective inspontaneous disease which further supports the aboveconclusions.10

Activation of Th1 or Th2 cells in response tomicrobial infection will depend upon many factors

Ž .including i the nature of cytokines produced by theinnate immune system in the host by the infection;Ž .ii cross-reactivity of microbial epitopes with islet b

Ž .cell antigens; and iii density and affinity of theseepitopes for MHC molecules. As shown in Figure 1A,viral infection of the islets could lead to the release ofself-antigens from the b cells. We have previouslyshown that cross-reactive epitopes with strong MHCbinding affinity induces the activation of Th1 respon-ses where as low affinity epitopes result in the activa-tion of Th2 responses of same antigen specificityŽ . 48Figure 1E . Microbial infection would result in thegeneration of epitopes with different affinities forMHC molecules some of which may cross-react with

Ž .islet antigens Figure 1B . This would result in theactivation of Th1 or Th2 cells. High viral dose orrepeated viral infection could result in Th1 responseswhile mild infection could result in Th2 responses

83

B. Singh et al

Ž .Figure 1E . Therefore the same virus could preventor induce IDDM in susceptible subjects. The directrole of Th1 and Th2 cells in IDDM remains unclearas both types of cells can transfer disease49. In sponta-neous disease, CD8 CTL are also required and theiractivation requires the Th1 cells.10

Immunization with bacterial preparations such asBCG, M. tuberculosis, OK432 are strongly protectivein IDDM.1 We were the first to demonstrate that asingle injection of mycobacterial preparation such asCFA or BCG at a young age prevented the develop-ment of IDDM in NOD mice and BB rats.50,51,52 BCGimmunization studies have also been carried out inhuman subjects soon after the disease diagnosis.53

Protection in NOD mice is associated with the pro-duction of protective cytokines such as IL-4,54,55 TGF-b and the down-regulation of ‘Fas-L’ expression in

Ž .the islets of treated animals unpublished results . Wehave also found that BCG immunization induces GAD

56 Žcross-reactive T cells of Th2 phenotype Qin et al.in preparation . There appears to be a window of

opportunity to influence the course of disease. Thiswindow is much narrower in BB rats than in NODmice. We found that the best protection was achievedin NOD mice if CFA was given between 4 and 10weeks of age, whereas in BB rats, adjuvant had to begiven before 4 weeks of age for best results.51,52 Werecently showed that cyclophosphamide accelerateddisease was also prevented by BCG immunization ifgiven within 3 days of treatment57. We have alsoshown that BCG also prevents the recurrence ofdisease in the islet transplanted diabetic NOD mice.58

Implications for the therapy of IDDM

The above studies predict a strong pathogenic orpreventative role for microbial agents in autoimmunediseases which may be exploited in therapeutic strate-gies. Vaccination against enteroviral infections may bean important step to prevent the development ofIDDM.14,16 Many bacterial agents such as BCG vac-cine offer a rational approach for the prevention ofautoimmune diseases.1 However, much work needsto be done to determine the dose, time and mode ofadministration of these therapies.59,60 Oral or in-tranasal administration of autoantigens such as in-sulin and GAD or their immunodominant epitopes inconjugation with microbial agents may offer powerfulnew approaches to prevent autoimmune diseasesthrough the mucosal immune system61. The role ofcross-reactive dietary antigens in IDDM also remains

an area of active investigation. We believe that mi-crobial products or autoantigens capable of inducingimmunoregulatory responses may provide alterna-tives to immune suppression for prevention of IDDMin subjects at high risk for developing disease. Thisapproach may also be useful for the prevention ofrecurrence of autoimmunity in islet transplantation.

Acknowledgements

The work of the authors was supported by the JuvenileDiabetes Foundation Medical Research Council of Canada,London Health Sciences Centre and the Canadian Dia-betes Association. We thank Dr Peter MacKay for helpfuldiscussion.

References

Ž .1. Singh B, Rabinovitch A 1993 Influence of microbial agentson the development of autoimmune diabetes. Autoimmunity15:209]213

Ž .2. Fohlman J, Friman G 1993 Is juvenile diabetes a viraldisease? Ann Med 25:569]574

Ž .3. Yoon JW 1990 The role of viruses and environmental factorsin the induction of diabetes. Curr Top Microbiol Immunol164:95]123

Ž .4. Maclaren NK, Atkinson MA 1997 Insulin-dependent dia-betes mellitus: the hypothesis of molecular mimicry betweenislet cell antigens and microorganisms. Mol Med Today3:76]83

Ž .5. Todd JA 1992 Genetic analysis of a susceptibility to type Idiabetes. Springer Semin Immunolopathol 14:33]58

Ž .6. Wicker LS, Todd JA, Peterson LB 1995 Genetic control ofautoimmune diabetes in the NOD mouse. Annu Rev Im-munol 13:179]200

Ž .7. Leiter EH 1990 The role of environmental factors in modu-lating insulin dependent diabetes, in The Role of Microor-

Žganisms in Non-infectious disease de Vries RRP, Cohen IR,.van Rood JJ, eds , pp 39]55. Springer Verlag, Berlin

8. Karges W, Hammond-McKibben D, Cheung RK, Viscanti M,Ž .Shibuya N, Kemp D, Dosch DH 1997 Immunological aspects

of nutritional diabetes prevention in NOD mice: a pilot studyfor the cow’s milk-based IDDM prevention trial. Diabetes46:557]664

9. Scott FW, Cloutiert HE, Kleemann R, Woerz-Pagenstert U,Ž .Rowsell P, Modler HW, Kolb H 1997 Potential mechanism

by which certain food promote or inhibit the development ofspontaneous diabetes in BB rats: dosage, timing, early effecton islet area, and switch in filtrate from Th1 to Th2 cells.Diabetes 46:589]598

Ž .10. Bach J-F 1994 Insulin-dependent diabetes mellitus as anautoimmune disease. Endocr Rev 15:516]542

Ž .11. Pozzilli P, Signore A, Williams AJ, Beales PE 1993 NODmouse colonies around the world]recent facts and figures.Immunol Today 14:193]6

Ž .12. Wucherpfennig KW, Strominger JL 1995 Molecular mimicryin T cell-mediated autoimmunity: viral peptides activate hu-man T cell clones specific for myelin basic protein. Cell80:695]705

13. Helfand RF, Gary HE Jr, Freeman CY, Anderson LJ, PallanschŽ .MA 1995 Serologic evidence of an association between

84

Microbial infection and IDDM

enteroviruses and the onset of type 1 diabetes mellitus. Pitts-burgh Diabetes Res Group. J Infect Dis 172:1206]11

14. Graves PM, Norris JM, Pallansch MA, Gerling IC, Rewers MŽ .1997 The role of enteroviral infections in the developmentof IDDM. Diabetes 46:161]168

Ž .15. Luppi P, Trucco M 1996 Superantigens in insulin-depen-dent diabetes mellitus. Springer Semin Immunopathol17:333]62

16. Dyrberg T, MacKay P, Michelsen B, Petersen J, Karlsen A,Ž .Bonnevie V 1996 Virus and Diabetes Mellitus, in Microor-

Žganisms and Autoimmune diseases Friedman H, Rose NR,.Bendinelli M, eds , pp 105]127. Plenum Press, New York

17. Buesa-Gomez J, de la Torre JC, Dyrberg T, Landin-Olssan M,Ž .Mauseth RS, Lernmark A, Oldstan MB 1994 Failure to

detect genomic viral sequences in pancreatic tissue from twochildern with acute-onset diabetes mellitus. J Med Virol42:193]197

18. Guberski DL, Thomas VA, Shek WR, Like AA, Handler ES,Ž .Rossini AA, Wallace JE, Welsh RM 1991 Induction of type I

diabetes by Kilham’s rat virus in diabetes-resistant BBrWorrats. Science 254:1010]1013

Ž .19. Yoon J-W 1991 Role of viruses in the pathogenesis of IDDM.Ann Med 23:437]45

Ž .20. Rabinovitch A 1994 Immunoregulatory and cytokine im-balances in the pathogenesis of IDDM. Therapeutic interven-tion by immunostimulation? Diabetes 43:613]21

Ž .21. Zechel M, Chaturvedi P, Singh B 1997 Characetrization ofimmunodominant peptide determinants of IDDM-associatedautoantigens in the NOD mice. Res Immunol 148:338]348

Ž .22. Rider BJ, Fraga E, Yu Q, Singh B 1996 Immune responses toself peptides naturally presented by MHC class II molecules.Mol Immunol 33:625]633

Ž .23. Agrawal B, Manickasundari M, Fraga E, Singh B 1991 Tcells exist in syngeneic mice which recognize peptide se-quences of self MHC Class II molecules. J Immunol147:383]390

Ž .24. Chaturvedi P, Qin H-Y, Chou H, Singh B 1997 ModulationŽ .of Insulin dependent diabetes mellitus IDDM in NOD mice

by autoreactive T cells. Crit Rev Immunol 17:519]52825. Gautam AM, Pearson CI, Lock CB, Steinman L, Smilek DE,

Ž .McDevitt HO 1994 Minimum structural requirements forpeptide presentation by major histocompatibility complexclass II molecules: implications in induction of autoimmunity.Proc Natl Acad Sci USA 91:767]71

26. Tisch R, Yang X-D, SInger SM, Liblau RS, Fugger L, McDevittŽ .HO 1993 Immune response to glutamic acid decarboxylase

correlates with insulitis in non-obese diabetic mice. Nature366:72]75

27. Kaufman DL, Clare SM, Tian J, Forsthuber T, Ting GS,Robinson P, Atkinson MA, Sercarz EF, Tobin AJ, Lehmann

Ž .PV 1993 Spontaneous loss of T-cell tolerance to glutamicwacid decarboxylase in murine insulin-dependent diabetes see

xcomments . Nature 366:69]72Ž .28. Karounos DG, Wolinsky JS, Thomas JW 1993 Monoclonal

antibody to rubella virus capsid protein recognizes a beta-cellantigen. J Immunol 150:3080]3085

29. Atkinson MA, Bowman MA, Campbell L, Darrow B, Kauf-Ž .mann DL, Maclaren NK 1994 Cellular immunity to a de-

terminant common to glutamate decarboxylase and Cox-sackie virus in insulin-dependet diabetes mellitus. J Clin In-vest 94:2125]2129

Ž .30. Tian J, Lehmann PV, Kaufmann DL 1994 T cell reactivitybetween Coxsackie virus and glutamate decarboxylase is asso-ciated with a murine diabetes susceptibility allele. J Exp Med180:1979]1984

31. Lehmann PV, Sercarz EE, Forsthuber T, Dayan CM, Gam-Ž .mom G 1993 Determinant spreading and the dynamics of

the autoimmune T-cell repertoire. Immunol Today14:203]208

Ž .32. Kappler JW, Roehm N, Marrack P 1987 T cell tolerance byclonal elimination in the thymus. Cell 49:273]280

33. Jardetzky TS, Brown JH, Gorga JC, Stern LJ, Urban RG, ChiŽ .YI, Stauffacher C, Straniger JL, Wiley DC 1994 Three-di-

mensional structure of a human class II histocompatibilitymolecule complexed with superantigen. Nature 368:711]718

Ž .34. Acha-Orbea H 1996 Superantigens of mouse mammarytumour virus: implications for diabetes? Diabetes Metab Rev12:111]119

Ž .35. Lafon M, Lafage M, Martinez-Arends A et al 1992 Evidencein humans of a viral superantigen. Nature 358:507]509

Ž .36. Conrad B, Weldmann E, Trucco G et al 1994 Evidence for asuperantigen involvement in insulin-dependent diabetes mel-litus aetiology. Nature 371:351]355

37. Conrad B, Weissmahr NR, Boni J, Arcari R, Schupbach J,¨ ¨Ž .Mach B 1997 A human endogenous retroviral superantigen

as candidate autoimmune gene in type I diatetes. Cell90:303]313

Ž .38. Kawamura T, Nagata M, Utsugi T, Yoon JW 1993 Preventionof autoimmune type I diabetes by CD4q suppressor T cellsin superantigen-treated non-obese diabetic mice. J Immunol151:4362]70

39. Shimada A, Taylor-Edwards C, Charlton B, Fathman CGŽ .1996 Beta-cell destruction may be a late consequence of theautoimmune process in nonobese diabetic mice. Diabetes45:1063]1067

Ž .40. Gazda LS, Charlton B, Lafferty KJ 1997 Diabetes resultsfrom a late change in the autoimmune response of NODmice. J Autoimmunity 10:261]270

Ž .41. Singh B, Dillon T, Fraga E, Lauzon J 1990 Role of the firstexternal domain of I-A beta chain in immune responses and

Ž .diabetes in non-obese diabetic NOD mice. J Autoimmunity3:507]21

Ž .42. Todd JA, Bell JI, McDevitt HO 1987 HLA-DQ b gene con-tributes to susceptibility and resistance to insulin-dependentdiabetes mellitus. Nature 329:599]604

43. Hanson MS, Cetkovic-Cvrlje M, Ramiya VK, Atkinson MA,Maclaren NK, Singh B, Elliott JF, Serreze DV, Leiter EHŽ .1996 Quantitative thresholds of MHC class II I-E expressedon hemopoietically derived antigen-presenting cells in trans-genic NODrLt mice determine level of diabetes resistanceand indicate mechanism of protection. J Immunol157:1279]87

Ž .44. Nepom GT, Erlich H 1991 MHC class-II molecules andautoimmunity. Annu Rev Immunol 9:493]525

45. Deng H, Apple R, Clare SM, Trembleau S, Mathis D, AdoriniŽ .L, Sercarz E 1993 Determinant capture as a possible mecha-

nism of protection afforded by major histocompatibility com-plex class II molecules in autoimmune disease. J Exp Med178:1675]1680

Ž .46. Abbas AK, Sher A, Murphy KM 1996 Functional diversity ofhelper T lymphocytes. Nature 383:787]793

Ž .47. Shevach E 1997 Microbial products induce autoimmunedisease by an IL-12-dependent pathway. J Immunol158:5087]5090

Ž .48. Chaturvedi P, Yu Q, Southwood S, Sette A, Singh B 1996Peptide analogs with different affinities for MHC alter thecytokine profile of T helper cells. Int Immunol 8:745]755

Ž . Ž .49. Pakala SV, Kurrer MO, Katz JD 1997 T helper 2 Th2 Tcells induce acute pancreatitis and diabetes in immune-com-

Ž .promised nonobese diabetic NOD Mice. J Exp Med186:299]306

Ž .50. Sadelain MW, Qin HY, Lauzon J, Singh B 1990 Preventionof type I diabetes in NOD mice by adjuvant immunotherapy.Diabetes 39:583]589

51. Sadelain MWJ, Qin HY, Samoski W, Parfrey N, Singh B,Ž .Rabinovitch A 1990 Prevention of diabetes in the BB rat by

early immunotherapy using Freund’s adjuvant. J Autoimmu-nity 3:671]680

85

B. Singh et al

Ž .52. Qin HY, Sadelain MW, Hitchon C, Lauzon J, Singh B 1993Complete Freund’s adjuvant-induced T cells prevent the de-velopment and adoptive transfer of diabetes in nonobesediabetic mice. J Immunol 150:2072]2080

53. Shehadeh NN, Calcinaro F, Bradley B, Bruchlim I, Vardi P,Ž .Lafferty KJ 1994 Effect of adjuvant therapy on development

of diabetes in mouse and man. Lancet 343:706]707Ž .54. Shehadeh NN, LaRosa F, Lafferty KJ 1993 Altered cytokine

activity in adjuvant inhibition of autoimmune diabetes. JAutoimmun 6:291]300

55. Rabinovitch A, Suarez-Pinzon WL, Sorensen O, Bleackley RC,Ž .Power RF 1996 IFN-g gene expression in pancretic islet-in-

filtrating mononuclear cells correlates with autoimmune dia-betes in nonobese diabetic mice. J Immunol 154:4874]4882

Ž .56. Qin HY, Elliott JF, Rajotte RV, Singh B 1996 EndogenousŽ .immune responses to glutamic acid decarboxylase GAD67

in NOD mice are modulated by adjuvant immunotherapy.Ž .Autoimmunity 24 suppl 1 :37

Ž .57. Qin HY, Singh B 1997 BCG vaccination prevents insulin-de-Ž .pendent diabetes mellitus IDDM in NOD mice after disease

accleration with cyclophosphamide. J Autoimmunity10:271]278

Ž .58. Lakey JR, Singh B, Warnock GL, Rajotte RV 1994 BCGimmunotherapy prevents recurrence of diabetes in islet graftstransplanted into spontaneously diabetic NOD mice. Trans-plantation 57:1213]1217

Ž .59. Singh B, Elliott JF 1997 Can progression of IDDM beprevented in newly diagnosed patients by BCG immunother-apy? Diabetes Prev Ther 11:12]13

Ž .60. Singh B 1997 Possible immunological treatment for Type IŽ .diabetes in the 21st century. Pract Diabetes Int 14 in press

61. Ma SW, Zhao DL, Yin ZQ, Mukherjee R, Singh B, Qin HY,Ž .Stiller CR, Jevnikar AM 1997 Transgenic plants expressing

autoantigens fed to mice to induce oral tolerance. NatureMedicine 3:793]796

86