TNF-alpha has been linked to the development of type 1 diabetes (T1D). We previously reported that neonatal treatment of nonobese diabetic (NOD) mice with TNF-alpha accelerated the onset of T1D, whereas TNF-alpha blockade in the same time period resulted in a complete absence of diabetes. The mechanisms by which TNF-alpha modulates development of T1D in NOD mice remain unclear. Here we tested the effects of TNF-alpha on the maturation of dendritic cells (DCs) in the NOD mouse. We found that neonatal treatment with TNF-alpha caused an increase in expression of maturation markers on CD11c(+)CD11b(+) DC subpopulations, whereas treatment with anti-TNF-alpha resulted in a decrease in expression of maturation markers in the CD11c(+)CD11b(+) subset. Moreover, neonatal treatment with TNF-alpha resulted in skewed development of a CD8alpha(+)CD11b(-)CD11c(+) DC subset such that TNF-alpha decreases the CD8alpha(+)CD11c(+) DC subset, increases the CD11c(+)CD11b(+) subset, and causes an increase in the expression of CD40 and CD54 on mature DCs capable of inducing immunity. Anti-TNF-alpha-treated mice had an increase in the CD8alpha(+)CD11c(+) DCs. Notably, adoptively transferred naïve CD4(+) T cells from BDC2.5 T cell receptor transgenic mice proliferated in the pancreatic lymph nodes in TNF-alpha-treated NOD mice but not in anti-TNF-alpha-treated mice. Finally, we show that anti-TNF-alpha-treated mice showed immunological tolerance to islet cell proteins. We conclude that TNF-alpha plays an important role in the initiation of T1D in the NOD mouse by regulating the maturation of DCs and, thus, the activation of islet-specific pancreatic lymph node T cells.

This presentation is an overview of mechanisms for developing and maintaining self-tolerance in mammalian organisms. Because this meeting is focused on type 1 diabetes and its mechanisms, the discussion deals primarily with mechanisms of T-cell tolerance, since type 1 diabetes in both effector and initiator phases is primarily a T-cell-mediated autoimmune disease. Emphasis is placed on more recently discovered mechanisms of maintaining self-tolerance (autoimmune regulator [AIRE]) and a new defect in T-cell negative selection. The emerging picture is that of a polygenic disease with various combinations of different alleles of many genes with important roles in the normal immune response or normal immune responses.

Glutamic acid decarboxylase (GAD) 65 is one of the major pancreatic antigens targeted by self-reactive T cells in type I diabetes mellitus. T cells specific for GAD65 are among the first to enter inflamed islets and may be important for the initiation of autoimmune diabetes. However, we previously reported that nonobese diabetic (NOD) mice transgenic for a T cell antigen receptor (TCR) specific for one of the immunodominant epitopes of GAD65, peptide 286-300 (G286), are protected from insulitis and diabetes. To examine whether other GAD65-reactive T cells share this phenotype, we have generated TCR transgenic NOD mice for a second immunodominant epitope of GAD65, peptide 206-220 (G206). As in G286 mice, G206 mice do not develop islet inflammation or diabetes. When adoptively transferred along with diabetogenic T cells, activated G206 T cells significantly delayed the onset of diabetes in NOD.scid recipients. Both G206 and G286 T cells produce immunoregulatory cytokines IFN-gamma and IL-10 at low levels when activated by cognate antigens. These data suggest that GAD65-specific T cells may play a protective role in diabetes pathogenesis by regulating pathogenic T cell responses. A better understanding of the functions of autoreactive T cells in type I diabetes will be necessary for choosing desirable targets for immunotherapy.

We previously described the generation of non- obese diabetic ( NOD) mice expressing a transgenic T cell receptor ( TCR) specific for peptide epitope 286 - 300 of the diabetes related self antigen, glutamic acid decarboxylase ( GAD) 65 in the context of I- A (g7) class II MHC, that are paradoxically protected from diabetes. In this report, we examine the atypical CD8(+) cells in these mice. Unlike typical class II restricted TCR transgenic mice, GAD286 mice have normal numbers of CD8(+) cells, half of which express high levels of the transgenic TCR. These MHC mismatched CD8(+) cells persist in the periphery and proliferate to GAD286 - 300 peptide in vitro and in vivo in a class II restricted fashion. Interestingly, the CD8(+) tetramer(-) T cells that are expressing endogenous TCR can delay diabetes induction in a transfer model, as we previously showed for CD4(+) tetramer(+) T cells in these mice. The MHC mismatched CD8(+) cells appear to be positively selected in an atypical fashion, in that they do not upregulate CD69 or reexpress CD44, and they escape negative selection. We find that production of these CD8(+) cells is not dependent on NOD thymus or high affinity of the TCR, but is dependent on the atypical TCR transgenic thymic environment.

The mechanism by which tumor necrosis factor-α (TNF) differentially modulates type I diabetes mellitus in the nonobese diabetic (NOD) mouse is not well understood. CD4+CD25+T cells have been implicated as mediators of self-tolerance. We show (i) NOD mice have a relative deficiency of CD4+CD25+T cells in thymus and spleen; (ii) administration of TNF or anti-TNF to NOD mice can modulate levels of this population consistent with their observed differential age-dependent effects on diabetes in the NOD mouse; (iii) CD4+CD25+T cells from NOD mice treated neonatally with TNF show compromised effector function in a transfer system, whereas those treated neonatally with anti-TNF show no alteration in ability to prevent diabetes; and (iv) repeated injection of CD4+CD25+T cells into neonatal NOD mice delays diabetes onset for as long as supplementation occurred. These data suggest that alterations in the number and function of CD4+CD25+T cells may be one mechanism by which TNF and anti-TNF modulate type I diabetes mellitus in NOD mice.

It has mystified immunologists for decades that antigen-specific T-cell clones obtained from human peripheral blood are, almost without exception, restricted by MHC class II molecules of the HLA-DR type, rather than by HLA-DQ molecules (1, 2). This is particularly surprising in that HLA-DQ alleles strongly influence susceptibility to many autoimmune diseases, including type 1 diabetes mellitus (T1DM) (3, 4). Thus, in humans, the HLA-DQB*0602 and HLA-DRB1*0403 alleles confer strong protection against T1DM (4, 5). The basis of this protective effect is undoubtedly complex and has proved difficult to study in humans. Unlike mice, where a number of inbred laboratory strains lack expression of H-2 IE (the murine equivalent to HLA-DR), humans always express both HLA-DR and HLA-DQ, usually a different set from each parent. In humans, it is therefore difficult to study the function of a single HLA class II allele in isolation, and in vivo experimentation with human subjects has many other constraints. These problems are further complicated by the strong linkage disequilibrium in the MHC class II region, which makes it difficult to distinguish the effects of individual HLA-DR alleles from the effects of linked HLA-DQ alleles, since particular DR/DQ allelic combinations tend to persist, with very little recombination. For these reasons, several laboratories have developed HLA class II transgenic mice to serve as an experimental surrogate for the CD4+ T-cell compartment of normal human subjects (6). David and his colleagues have produced HLA class II transgenic mice, which they and others have used to model human autoimmune diseases (7–9). By studying the influence of different combinations of coexpression of DR3, DQ8, DR2, and/or DQ6 alleles, and comparing these with HLA transgenic mice carrying only a single HLA allele, these authors found that coexpression of DR3 can modify the manifestations of an experimental arthritis associated with expression of DQ8 (7). DQ8 also increases the severity of other experimentally induced autoimmune diseases in this model system (8), whereas the DQ6 allele can prevent the spontaneous loss of tolerance to the pancreatic autoantigen GAD65 and the insulitis that are otherwise seen in HLA-DR3, DQ8 transgenic mice (9). In this issue of the JCI, Wen et al. wished to explore the influence of coexpression of HLA-DR alleles on the diabetes susceptibility conferred by the DQ8 allele (10). The HLA-DRB1*0401, DQ8 haplotype, the most common HLA haplotype in Caucasian T1DM patients, was therefore a logical candidate for this study (5). Because HLA-DQ8 transgenic mice do not develop diabetes spontaneously, the authors crossed them with a diabetes-prone transgenic strain, RIP-B7, which lacks endogenous MHC class II molecules and overexpresses the costimulatory molecule B7 specifically in the islet cells of the pancreas (11). When these RIP-B7 transgenic animals also carry the human HLA-DQ8 transgene, they develop spontaneous diabetes at a frequency of 73% (10, 11). Remarkably, a congenic RIP-B7 transgenic line carrying the DR4 (B1*0401) allele develops diabetes at a frequency of only 25%. To model the case of humans expressing the HLA-DR4, DQ8 haplotype, the authors then produced mice expressing both DR4 and DQ8 and found that the incidence of disease in these HLA double transgenic mice is almost identical to that of the DR4 transgenic mice (10). This suggests that DR4 can partly cancel the disease-promoting effect of DQ8. Following up on this finding, Wen and colleagues purified splenic CD4+ T cells from the diabetes-prone DQ8/RIP-B7 animals and from the other, relatively disease-insensitive strains (10). In vitro cytokine production by cells from DQ8/RIP-B7 mice produced mainly IFN-γ, compatible with a Th1 cytokine pattern. In contrast, cells from the DR4/RIP-B7 and the DQ8DR4/RIP-B7 mice produced IL-4 but very little IFN-γ, suggesting a Th2 cytokine pattern (10). However as the authors point out, the mechanisms of the HLA-DR4 effect are not clear (10). Similar HLA class II transgenic mice have been used to study the T-cell receptor (TCR) repertoire after immunization with recombinant GAD65 (12). These studies, using several DR/DR as well as DR/DQ combinations, have revealed that certain immunodominant GAD65-specific TCRs are all but absent in the repertoire of HLA class II double transgenic mice, if the diabetes-protective DRB1*0403 allele is present. This elimination of potentially pathogenic T cells occurs through still-unidentified mechanisms, operating at least in part at the level of intracellular antigen processing. Similarly, human antigen-presenting cells (APCs) carrying both the diabetes-protective DRB1*0403 allele and a second, diabetes-conducive DR4 allele (DRB1*0405) fail to present these processed immunodominant GAD65 epitopes to T cells (S. Parry and G. Sonderstrup, unpublished results), although they are perfectly capable of presenting these epitopes if they are provided as synthetic peptides. Further, these same human APCs can process and present other GAD65 epitopes normally to T cells that recognize these epitopes in the context of the DRB1*0405 molecule (S. Parry and G. Sonderstrup, unpublished results). This result indicates that this is a selective epitope-specific phenomenon. The ability to eliminate potentially pathogenic T cells presented by coexpressed HLA class II molecules, HLA-DR or DQ, may be responsible for the disease-protective effect of the HLA-DRB1*0403 allele and may therefore represent yet another mechanism for preventing autoimmunity in humans. The study of Wen et al. (10) and the results described above illustrate how HLA class II transgenic mice can provide a blueprint for unraveling DR/DQ allelic interactions in autoimmunity. This information should also provide new insight into pathogenesis and inspire novel strategies to prevent autoimmune disease.

To assess the role of lymphotoxin-β receptor (LTβR) in diabetes pathogenesis, we expressed an LTβR–Fc fusion protein in nonobese diabetic (NOD) mice. The fusion protein was expressed in the embryo, reached high levels for the first 2 wk after birth, and then declined progressively with age. High expression of LTβR–Fc blocked diabetes development but not insulitis. After the decline in chimeric protein concentration, mice became diabetic with kinetics similar to the controls. Early expression of fusion protein resulted in disrupted splenic architecture. However, primary follicles and follicular dendritic cells, but not marginal zones, developed in aged mice. Hence, LTβR signaling is required for diabetes development and regulates follicular and marginal zone structures via qualitatively or quantitatively distinct mechanisms.

The discovery that genes in the major histocompatibility complex (MHC) play an important role in the immune response depended on the chance interaction of several unrelated events. The first, and most important, was the decision by Michael Sela to synthesize a series of branched, multichain, synthetic polypeptides based on a backbone of poly-l-lysine. The prototype compound, (T,G)-A-L, was tipped with short random sequences of tyrosine and glutamic acid. This resulted in a restricted range of antigenic determinants composed of only two or three amino acids with a variable length-ideal for binding to the peptide binding groove of MHC class II molecules. The second was the decision by John Humphrey to immunize various strains of rabbits with this synthetic polypeptide. Two of these rabbit strains showed very large quantitative differences in antibody response to (T, G)-A-L. In transferring this system to inbred mouse strains, the third bit of good fortune was the availability at the National Institute of Medical Research, in Mill Hill (London), of the CBA (H2(k)) and C57 (H2(b)) strains. The H2(b) haplotype is the only one mediating a uniform high antibody response to (T,G)-A-L. The fourth critical ingredient was the availability of numerous congenic and H2 recombinant inbred strains of mice produced earlier by Snell, Stimpfling, Shreffler, and Klein. A search for congenic pairs of mice expressing the responder and nonresponder H2 haplotypes on the same background revealed that these strains responded as a function of their H2 haplotype, not of their inbred background. Extensive studies in a variety of inbred strains carrying recombinant H2 haplotypes, as well as a four-point linkage cross, mapped immune response to (T,G)A-L within the murine MHC, between the K and Ss loci. The demonstration that stimulation in the mixed lymphocyte reaction (MLR) mapped to the same region quickly led to attempts to produce antisera in congenic H2 recombinant strain combinations. These antisera identified I-region associated (Ia) antigens. Immunoprecipitation and blocking studies showed that the gene products controlling specific immune responses, the mixed lymphocyte reaction, and the structure of Ia antigens were one and the same-now designated as the I-A MHC class II molecules. These antisera and inbred strains enabled Unanue to demonstrate the peptide binding function of class II MHC molecules.

Particular HLA class II allelic sequences are associated with susceptibility to type I diabetes. To understand the mechanism, knowledge of the molecular nature of the specific TCR/peptide/class II interactions involved in the disease process is required. To this end, we have introduced the diabetes-associated human class II HLA-DQ8 allele (DQA1*0301/DQB1*0302) as a transgene into mice and analyzed T cell responses restricted by this molecule to an important Ag in human diabetes, human glutamic acid decarboxylase 65. Hybridomas were used to determine the particular peptides from this Ag presented by HLA-DQ8 to T cells and to map the core minimal epitopes required for T cell stimulation. Analysis of these core epitopes reveals a motif and relevant features for peptides that are immunogenic to T cells when presented by HLA-DQ8. The major immunogenic epitopes of glutamic acid decarboxylase 65 do not contain a negatively charged residue that binds in the P9 pocket of the HLA-DQ8 molecule. PBMC from HLA-DQ8+ diabetic and nondiabetic individuals respond to these peptides, confirming that the mouse model is a useful tool to define epitopes of autoantigens that are processed by human APC and recognized by human T cells.

Summary: This review examines the field of current HLA class II transgenic mouse models and the individual approaches applied in production of these mice. The majority of these mice have been created with the objective of obtaining a disease model with clinical features mimicking human autoimmune disease. The development process of a different type of HLA class II transgenic mice, which are designed to function as a substitute for a normal human immune system in studies of human autoantigens, is described. Several HLA‐DR4 transgenic lines with normally expressed HLA‐DR4 molecules have been produced. To obtain adequate positive selection of the HLA‐DR4‐restricted CD4 + T‐cell repertoire in these mice it is essential both to introduce a human CD4 transgene. and to delete the murine major histocompatibility complex (MHC) class II molecules. These HLA‐DR4 transgenic mice have been used to determine the immunogenic CD4 + T‐cell epitopes of several human autoantigenic proteins.

The N-terminal peptide Ac1-11 of myelin basic protein induces experimental autoimmune encephalomyelitis in H-2(u) and (H-2(u) x H-2(s)) mice but does not in H-2(s) mice. Ac1-11 binds weakly to the class II major histocompatibility complex (MHC) molecule I-Au but not at all to I-As. We have studied the interaction of Ac1-11 and I-Au as a model system for therapeutic intervention in the autoimmune response seen in experimental autoimmune encephalomyelitis. Two polymorphic residues that differ between I-Au and I-As, Y26beta and T28beta, and one conserved residue, E74beta, confer specific binding of Ac1-11 to I-Au. A fourth residue, R70beta in I-Au, affects both peptide binding and T cell recognition. These results are consistent with a model that places arginine at position five of Ac1-11 in pockets 4 and 7 of the MHC groove, which is formed in part by residues 26, 28, 70, and 74 of Abetau and places lysine at position four of Ac1-11, previously shown to be a major MHC contact, in hydrophobic pocket 6. The data indicate that the primary region of I-Au that confers specific binding of Ac1-11 lies in the center of the peptide binding groove rather than in the region that contacts the N terminus of the peptide, as has been shown for HLA DR and the homologous I-E molecules.

Susceptibility to type I diabetes is linked to class II MHC alleles in both mouse and man. However, the molecular mechanisms by which MHC molecules mediate disease susceptibility are unknown. To analyze how I-A alleles predispose to, or prevent, the development of type I diabetes, we have chosen, as the first step, to investigate the immune response to an important islet cell protein in diabetes-susceptible and diabetes-resistant mice. MHC class II alleles conferring susceptibility and resistance to diabetes select completely different sets of immunogenic epitopes from the beta islet cell autoantigen glutamic acid decarboxylase 65. Peptide-binding studies, analysis of MHC restriction, and immunization with these peptide epitopes indicate that the two amino acid substitutions within the I-A(beta) chain that distinguish a diabetes-susceptibility from a diabetes-resistance allele are sufficient to alter peptide binding and MHC restriction and may also influence antigen presentation and the selection of the T cell repertoire. The data indicate that the molecular mechanisms for class II-mediated selection of immunodominant epitopes are complex and differ for each individual peptide epitope. Further study of the functional characteristics of the response to these epitopes should provide insight into mechanisms of MHC-mediated diabetes susceptibility.

Previously, we have reported that neutralization of surface lymphotoxin (LT-alphabeta) in mice which expressed an LT-beta receptor-Fc fusion protein, driven by the cytomegalovirus promoter, resulted in an array of anatomic abnormalities. We now report that mice which express a tumor necrosis factor (TNF) receptor p60-Fc fusion protein (which neutralizes TNF and soluble LT-alpha3 activity) develop unique lymphoid abnormalities. Our data demonstrate that some aspects of peripheral lymphoid organ development require both surface LT-alphabeta and TNF interacting with their specific receptors. However, these related cytokines are also capable of signaling distinct developmental events. Splenic MAdCAM-1 expression, follicular dendritic cell localization and normal Peyer's patch development all require both surface LT-alphabeta and TNF activity. Marginal zone formation and splenic B cell localization primarily require surface LT-alphabeta-LT-beta receptor interactions. Primary follicle formation was dependent upon TNF receptor(s) engagement. Interestingly spleen, lymph nodes and Peyer's patches from TNF receptor p60-Fc-expressing mice all develop different abnormalities, suggesting distinct pathways of development in these lymphoid organs. Thymus development appears to be independent of these signaling pathways. These results demonstrate that TNF and LT are crucial for normal peripheral, but not central lymphoid organ development.

Susceptibility to the human autoimmune disease IDDM is strongly associated with those haplotypes of the major histocompatibility complex (MHC) carrying DQB1 alleles that do not encode aspartic acid at codon 57. Similarly, in a spontaneous animal model of this disease, the NOD mouse, the genes of the MHC play an important role in the development of diabetes. The DQB1 homolog in NOD mice, I-Ab(g7), encodes a histidine at codon 56 and a serine at codon 57, while all other known I-Ab alleles encode proline and aspartic acid, respectively, at these positions. We therefore mutated the NOD I-Ab allele to encode proline at position 56 and aspartic acid at position 57 and introduced this allele onto the NOD genetic background to study the effect of these substitutions on susceptibility to diabetes. No transgenic mice developed diabetes by 8 months of age, and transgenic mice had markedly reduced lymphocytic infiltration in the pancreas compared with nontransgenic littermates. Furthermore, splenocytes from transgenic mice failed to proliferate or secrete gamma-interferon in response to a panel of beta-cell autoantigens, although the mice did produce beta-cell specific antibodies. Interestingly, the proportion of IgG1 and IgE relative to IgG2a comprising these autoantibodies was much greater in transgenic mice compared with nontransgenic control mice. Finally, T-cells from transgenic mice inhibited the adoptive transfer of diabetes to irradiated recipients. This inhibition was partially reversed by treatment of the recipients with a combination of anti-interleukin (IL)-4 and anti-IL-10 monoclonal antibodies. Thus, a transgenic class II MHC allele encoding aspartic acid at B57 prevents diabetes, in part, by promoting the production of IL-4 and IL-10, which interfere with the effector phase of the diabetic process.

IDDM is a T-cell-mediated autoimmune disease in which the insulin-producing beta-cells are destroyed. The disease process is complex, involving the recognition of several beta-cell autoantigens. One of these, GAD65, appears to have a critical and not fully defined role in IDDM in humans and in the NOD mouse. We provide evidence that an ongoing diabetogenic response in NOD mice can be suppressed after intravenous administration of GAD65, but not by other beta-cell autoantigens. Furthermore, suppression of the diabetogenic response is mediated by the induction of GAD65-specific CD4+ regulatory T-cells. Finally, cytokine analysis indicates that these CD4+ regulatory T-cells have a T-helper 2 phenotype.

The nonobese diabetic (NOD) mouse spontaneously develops T cell-dependent autoimmune diabetes. Here, we investigate the role of CD40 ligand (CD40L)-CD40 costimulation in the initiation and progression of this disease. Anti-CD40L mAb treatment of 3- to 4-wk-old NOD females (the age at which insulitis typically begins) completely prevented the insulitis and diabetes. In contrast, treatment of such mice with anti-CD40L at >9 wk of age did not inhibit the disease process. These results suggest that a costimulatory signal by CD40L is required early but not in the effector phase of disease development. Anti-CD40L treatment affected the priming of islet Ag-specific T cell responses in vivo. Cytokine analysis revealed a dramatic decrease in IFN-gamma and IL-2 release without a concomitant increase in IL-4 production by T cells from anti-CD40L-treated mice. Thus, anti-CD40L impaired the islet Ag-specific Th1 cell response in vivo, and the prevention of diabetes by anti-CD40L was not associated with switching of the response from a Th1 to a Th2 profile. Cotransfer of splenocytes from anti-CD40L-treated mice with splenocytes from diabetic NOD mice into NOD/scid mice did not inhibit the transfer of disease, indicating that anti-CD40L does not prevent the disease by inducing regulatory cells. Since anti-CD40L clearly prevented the insulitis by inhibiting the development and further accumulation of pathogenic Th1 cells to islets of Langerhans, we conclude that CD40L-CD40 costimulation is required for early events in the development of spontaneous autoimmune diabetes.

Experimental autoimmune encephalomyelitis (EAE) serves as a rodent model of the autoimmune disease multiple sclerosis. In mice, EAE is induced by immunizing with spinal cord homogenate, components of the myelin sheath, such as myelin basic protein (MBP) or proteolipid protein (PLP), or peptides derived from these components. EAE can be induced in H-2u or (H-2u × H-2s)F1 mice with the N-terminal peptide of MBP, Ac1-11. Coimmunization with Ac1-11 and Ac1-11[4A], an analog in which lysine at position four is substituted with alanine, prevents EAE. The mechanism of inhibition has not been elucidated, but probably does not work through MHC blockade, T cell anergy or clonal elimination of encephalitogenic T cells. We have isolated T cell clones and hybridomas from (PL/JXSJL/J)F1 mice immunized with either Ac1-11 alone or Ac1-11 and Ac1-11[4A] and analysed these cells for differences in their T cell receptor repertoire andin vitro response. Although T cells elicited by coinjection of Ac1-11 and Ac1-11[4A] expressed TCR that used Vα and Vβ gene elements similar to those elicited by Ac1-11 alone, they differed in the sequences of the junctional region of the alpha chain. Most of these T cells also responded less well to Ac1-11in vitro, suggesting that coinjection of Ac1-11 and Ac1-11[4A] preferentially activates T cells bearing TCR of different affinity for Ac1-11 bound to I-Au, and which may therefore be less encephalitogenic. Furthermore, our results show that a more diverse repertoire of Vα and Vβ genes are elicited by Ac1-11 in (PL/JXSJL/J)F1 mice compared to PL/J and B10.PL mice, providing further evidence that a restricted TCR repertoire is not required for the development of autoimmune disease.

Nonobese diabetic (NOD) mice develop autoimmune-mediated lymphocytic inflammation of pancreatic islets (insulitis) that leads to beta-cell destruction and development of diabetes. Inflamed islets show expression of lymphocyte alpha 4 beta 7 integrin and endothelial mucosal addressin cell adhesion molecule-1 (MAdCAM-1), adhesion molecules involved in tissue-selective migration of lymphocytes to mucosal lymphoid tissues. To elucidate the roles of the mucosal lymphocyte/endothelial adhesion system in the development of diabetes, we treated NOD mice with monoclonal antibody against beta 7 integrin or MAdCAM-1. Treatment of mice from age 7 to 28 days or 8 to 12 weeks with either antibody led to significant and long-standing protection against the spontaneous development of diabetes and insulitis. In contrast, neither treatment prevented the development of salivary gland inflammation (sialadenitis), indicating that the effect was tissue-selective. Monoclonal antibody treatment had no demonstrable effect on numbers or phenotypes of peripheral lymphocytes or on the immune response to pancreatic islet or exogenous antigens. These data indicate that lymphocyte and endothelial adhesion molecules involved in the migration of lymphocytes into mucosal lymphoid tissues play a role in the development of diabetes in NOD mice. Moreover, the results suggest that treatment of humans with antibodies against tissue-selective lymphocyte or endothelial adhesion molecules may selectively inhibit the development of autoimmune diseases such as diabetes.