Novel strategy for identification of candidate cytotoxic T-cell epitopes from human preproinsulin.

We describe a strategy for identifying ligands of human leukocyte antigen (HLA) class I molecules based on a peptide library-mediated in vitro assembly of recombinant class I molecules. We established a microscale class I assembly assay and used a capture ELISA to quantify the assembled HLA-peptide complexes. The identity of the bound ligands was then deduced by mass spectrometry. In this method, HLA complexes assembled in vitro in the presence of components of a mixture of peptides were immunoprecipitated and the bound peptide(s) identified by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry. This process of epitope extraction is robust and can be used with complex mixtures containing in excess of 300 candidate ligands. A library of overlapping peptides representing all potential octamers, nonamers and decamers from human preproinsulin was synthesized using unique library chemistry. Peptides from the library were used to initiate assembly of recombinant HLA-B8, HLA-B15 and HLA-A2, facilitating the identification of candidate T-cell epitopes from preproinsulin.

HLA genes associated with autoimmunity and progression to disease in type 1 diabetes.

Insulin dependent diabetes mellitus (type I DM) is caused by an autoimmune process which culminates in destruction of pancreatic beta cells with resultant loss of insulin production. Preceding the clinical diagnosis of type I DM is a preclinical stage characterized by autoantibodies to insulin, glutamic acid decarboxylase (GAD) and a tyrosine phosphatase-like molecule (IA-2). We have studied both HLA class I and class 2 allele distributions in diabetic probands and autoantibody positive individuals in members of 452 families recruited for the Australian type I diabetes DNA repository. The results demonstrate that progression to autoimmunity as measured by the appearance of autoantibodies is strongly associated with the class 2 alleles DRB1*03 and DRB*04 and with DRB1*03/04 heterozygosity. In contrast, the progression to clinical disease appears associated with class I alleles A24, A30 and B18 while A1, A28, B14 and B56 appear negatively associated. The class 2 alleles appear to have a minimal role in the progression from autoantibody positivity to clinical disease. These results are consistent with the view that CD4+ T cells responding to peptides in the context of class 2 molecules are responsible for initiating autoantibody production, while the destruction of islet cells leading to clinical expression of the disease is the function of CD8+ T cells recognizing relevant peptides in the context of class I molecules.

Cytotoxic T cells to an epitope in the islet autoantigen IA-2 are not disease-specific.

Cytotoxic CD8 T lymphocytes (CTL) are effectors of pancreatic islet beta-cell destruction in type 1 diabetes but, with the exception of a single report, CTL to islet antigen peptides have not been identified. We used autologous blood monocyte-derived dendritic cells to elicit HLA-A2-restricted CTL to a peptide, MVWESGCTV (aa 797-805), that is contiguous with a dominant CD4 T-cell epitope in the islet antigen tyrosine phosphatase IA-2. IA-2 peptide-specific CTL activity measured as 51Cr release from autologous lymphoblasts was detected in 2/6 islet antibody-positive relatives at high risk for type 1 diabetes but also in 2/6 closely HLA-matched controls. All subjects had CTL activity to an HLA-A2-restricted Epstein-Barr virus peptide. CTL to the IA-2 self-peptide were therefore not disease-specific, consistent with other evidence that autoreactive T cells are present in healthy individuals.

Association between rotavirus infection and pancreatic islet autoimmunity in children at risk of developing type 1 diabetes.

Pancreatic islet autoimmunity leading to type 1 diabetes could be triggered by viruses in genetically susceptible individuals. Rotavirus (RV), the most common cause of childhood gastroenteritis, contains peptide sequences highly similar to T-cell epitopes in the islet autoantigens GAD and tyrosine phosphatase IA-2 (IA-2), suggesting T-cells to RV could trigger islet autoimmunity by molecular mimicry. We therefore sought an association between RV infection and islet autoantibody markers in children at risk for diabetes who were followed from birth. There was a specific and highly significant association between RV seroconversion and increases in any of these antibodies: 86% of antibodies to IA-2, 62% to insulin, and 50% to GAD first appeared or increased with increases in RV IgG or IgA. RV infection may therefore trigger or exacerbate islet autoimmunity in genetically susceptible children.

Screening for preclinical type 1 diabetes in a discrete population with an apparent increased disease incidence.

Environmental agents are proposed to play a role in triggering or exacerbating pancreatic islet autoimmunity in people genetically predisposed to type 1 diabetes. However, with few exceptions, these agents remain enigmatic. Clues to environmental agents may be found by investigating population/geographic clusters or 'hotspots' of high disease incidence. We were alerted to a small community where the incidence of type 1 diabetes appeared to be five-fold higher than expected. Because type 1 diabetes is now recognized to have a subclinical phase during which anti-islet antibodies can be detected, we aimed to identify and characterize a reservoir of children with subclinical disease in this community. Venous blood samples were collected from 1906/2347 (81%) local school children during one week. Islet cell antibodies (ICAs) were detected in 122 (6.4%) children, 18 (0.9%) being high titer (> or = 20 Juvenile Diabetes Foundation units (JDFu)). On retest, 15 months later, the majority of low titer ICAs were undetectable, whereas high-titer ICAs persisted. The latter were found in two distinct age-related, ethnically similar groups. The younger group, aged 6-9 yr, had antibodies to insulin (IAAs), glutamic acid decarboxylase (GAD) and tyrosine phosphatase IA2 in addition to ICA, human leukocyte antigen (HLA) genes associated with susceptibility to type 1 diabetes, and lower first-phase insulin responses (FPIRs) to intravenous glucose. The older group, aged 13-16 yr, the age cohort of the index clinical cases, had few antibodies other than ICA, non-susceptibility HLA genes and normal FPIRs. During follow-up, three children, all from the younger group with multiple antibodies and FPIRs less than the first percentile, developed diabetes 4, 6 and 7 yr after screening. The finding of two age groups of subclinical disease suggests that if environmental agents triggered islet autoimmunity they did not act constantly on the community. Furthermore, the absence of multiple autoantibodies and/or HLA susceptibility genes in the older group, the source of index clinical cases, implies they are a residual subgroup with slow or absent progressive beta-cell destruction. This study illustrates that the natural history of type 1 diabetes may be elucidated by analyzing age-related subclinical disease in the general population.

Cow's milk and type 1 diabetes: the real debate is about mucosal immune function.

The hypothesis that early exposure of the infant to cow's milk (or lack of breast-feeding) predisposes the child to type 1 diabetes dates from the 1980s. It has important implications, but remains controversial because the evidence on which it is based has been indirect and is open to criticism. Two meta-analyses of multiple studies in which diabetes prevalence was associated retrospectively with infant feeding revealed only a marginal increase in relative risk. Two recent prospective studies found no apparent association between development of antibodies to islet antigens and feeding patterns in high-risk infants with a first-degree type 1 diabetic relative. Studies reporting increased humoral and cellular immunity to cow's milk proteins in children with type 1 diabetes often lack appropriate controls and standardization and do not, in themselves, establish a causal connection to disease pathogenesis. A review of published data leads to the conclusion that increased immunity to cow's milk proteins is not disease-specific, but reflects genetic predisposition to increased immunity to dietary proteins in general, associated with the HLA haplotype A1-B8-DR3-DQ2 (A1*0501, B1*0201), which also predisposes to celiac disease and selective IgA deficiency. We suggest that the cow's milk hypothesis could be productively reframed around mucosal immune function in type 1 diabetes. Breast milk contains growth factors, cytokines, and other immunomodulatory agents that promote functional maturation of intestinal mucosal tissues. In the NOD mouse model, environmental cleanliness may influence diabetes incidence through mucosal mechanisms, and exposure of the mucosa to insulin (present in breast milk) induces regulatory T-cells and decreases diabetes incidence. The mucosa is a major immunoregulatory barrier, and cow's milk happens to be the first dietary protein it encounters. The basic question is whether impaired mucosal immune function predisposes to type 1 diabetes.

Neural network-based prediction of candidate T-cell epitopes.

Activation of T cells requires recognition by T-cell receptors of specific peptides bound to major histocompatibility complex (MHC) molecules on the surface of either antigen-presenting or target cells. These peptides, T-cell epitopes, have potential therapeutic applications, such as for use as vaccines. Their identification, however, usually requires that multiple overlapping synthetic peptides encompassing a protein antigen be assayed, which in humans, is limited by volume of donor blood. T-cell epitopes are a subset of peptides that bind to MHC molecules. We use an artificial neural network (ANN) model trained to predict peptides that bind to the MHC class II molecule HLA-DR4(*0401). Binding prediction facilitates identification of T-cell epitopes in tyrosine phosphatase IA-2, an autoantigen in DR4-associated type1 diabetes. Synthetic peptides encompassing IA-2 were tested experimentally for DR4 binding and T-cell proliferation in humans at risk for diabetes. ANN-based binding prediction was sensitive and specific, and reduced the number of peptides required for T-cell assay by more than half, with only a minor loss of epitopes. This strategy could expedite identification of candidate T-cell epitopes in diverse diseases.

Impaired yield, phenotype, and function of monocyte-derived dendritic cells in humans at risk for insulin-dependent diabetes.

Dendritic cells (DC) present Ag to naive T cells and are therefore pivotal in shaping immune responses. DC may either immunize or tolerize T cells. Humans with pancreatic islet autoimmunity at high risk for insulin-dependent diabetes mellitus (IDDM) present the opportunity to investigate DC in autoimmune disease. We compared DC phenotype and function in 12 euglycemic, asymptomatic IDDM relatives with islet autoimmunity and controls matched for age, sex, and MHC class II alleles. DC were generated from adherent peripheral blood cells by culture with granulocyte/macrophage-CSF and IL-4. The yield of DC was significantly lower in IDDM relatives than in controls. While the DC phenotype, HLA-DR+CD14-, was expressed by > or =90% of the cells generated from relatives and controls, the proportion of cells that expressed CD1a and the costimulator molecules CD80 (B7-1) and CD86 (B7-2) was significantly lower in IDDM relatives. In addition, B7-1 and B7-2 expression per cell was significantly lower in IDDM relatives. These phenotypic changes were accompanied by reduced stimulation of autologous CD4 cells by DC from IDDM relatives. Similar findings were obtained in three recently diagnosed IDDM patients. These findings indicate that impairment of DC phenotype and function is a marker of islet autoimmunity and are consistent with a role for impaired DC function in the pathogenesis of autoimmune disease.

The Melbourne Pre-Diabetes Study: prediction of type 1 diabetes mellitus using antibody and metabolic testing.

OBJECTIVES: To determine the utility of various autoantibodies in predicting progression to clinical diabetes in first-degree relatives of patients with type 1 diabetes mellitus. PARTICIPANTS: 3315 first-degree relatives of patients with type 1 diabetes (1161 parents, 1206 siblings and 948 offspring) recruited through diabetes clinics, private endocrinologists, Diabetes Australia and the Juvenile Diabetes Foundation. MAIN OUTCOME MEASURES: Prevalence of islet cell antibodies (ICA) levels > or = 20 JDFu, insulin autoantibodies (IAA) levels > 100 nU/mL, and antibodies to glutamic acid decarboxylase (GADAb) and tyrosine phosphatase IA2 (IA2Ab); change in beta cell function over time; and development of clinical diabetes. RESULTS: 2.6% of relatives had elevated ICA levels, 1.3% had elevated IAA levels and 0.3% had both. High ICA levels were significantly more frequent in siblings than in offspring or parents, and were more frequent in relatives younger than 20 years. GADAb were detected in 68% and IA2Ab in 57% of relatives with elevated ICA and/or IAA levels. Diabetes developed in 33 relatives (25 siblings, 2 offspring and 6 parents). Before diagnosis of clinical diabetes, high ICA levels were detected in 18 (58%), high IAA levels in 7 (23%), both in 5 (15%), and either in 19 (61%); GADAb were detected in 26 (84%), IA2Ab in 13 (42%), both in 11 (35%), and either in 28 (90%). First phase insulin release (FPIR) less than 50 mU/L was very strongly associated with progression to diabetes. In relatives with FPIR initially greater than 50 mU/L who eventually developed diabetes, there was a gradual and continuous reduction in FPIR over time before diagnosis. CONCLUSIONS: Type 1 diabetes can be diagnosed in the preclinical stage. The recently described antibodies to glutamic acid decarboxylase and tyrosine phosphatase IA2 appear superior to ICA as screening tools for the preclinical diagnosis of type 1 diabetes.

IgG subclass antibodies to glutamic acid decarboxylase and risk for progression to clinical insulin-dependent diabetes.

Pancreatic islet beta cell destruction leading to insulin-dependent diabetes mellitus (IDDM) is believed to be mediated by a T-helper 1 (T(H)1) lymphocyte response to islet antigens. In the mouse, T(H)1 (IL-2, IFN-gamma) and T(H)2 (IL-4, -5, -6, -10) responses are associated with the generation of IgG2a and IgG1 subclasses, respectively. The equivalent human subclasses have not been defined. Because the IgG subclass response to an antigen may be a potentially useful marker of T(H)1/T(H)2 immune balance we measured IgG subclass antibodies to glutamic acid decarboxylase (GAD), a major islet autoantigen in IDDM, in 34 newly-diagnosed IDDM patients and in 28 at-risk, first-degree relatives of people with IDDM. In the newly-diagnosed patients, total IgG antibodies to GAD were detected in 74% (25/34); IgG1 and/or IgG3 were significantly more frequent than IgG4 or IgG4/IgG2 (14/34 versus 5/34, p = 0.01). GAD antibody-negative patients were significantly younger (p = 0.01). In 15 at-risk relatives who had not progressed to clinical diabetes after a median of 4.5 years, 10 had IgG2 and/or IgG4 antibodies compared to only 3/13 progressors (p = 0.02). Total IgG and IgG2 antibodies were higher in non-progressors. Non-progressors were older than progressors (p = 0.01), and relatives with IgG2 and/or IgG4 responses were also older (p = 0.01). These results suggest that IgG subclass antibodies to GAD may contribute to diabetes risk assessment in islet antibody relatives.

T-cell epitopes in type 1 diabetes autoantigen tyrosine phosphatase IA-2: potential for mimicry with rotavirus and other environmental agents.

The tyrosine phosphatase IA-2 is a molecular target of pancreatic islet autoimmunity in type 1 diabetes. T-cell epitope peptides in autoantigens have potential diagnostic and therapeutic applications, and they may hold clues to environmental agents with similar sequences that could trigger or exacerbate autoimmune disease. We identified 13 epitope peptides in IA-2 by measuring peripheral blood T-cell proliferation to 68 overlapping, synthetic peptides encompassing the intracytoplasmic domain of IA-2 in six at-risk type 1 diabetes relatives selected for HLA susceptibility haplotypes. The dominant epitope, VIVMLTPLVEDGVKQC (aa 805-820), which elicited the highest T-cell responses in all at-risk relatives, has 56% identity and 100% similarity over 9 amino acids (aa) with a sequence in VP7, a major immunogenic protein of human rotavirus. Both peptides bind to HLA-DR4(*0401) and are deduced to present identical aa to the T-cell receptor. The contiguous sequence of VP7 has 75% identity and 92% similarity over 12 aa with a known T-cell epitope in glutamic acid decarboxylase (GAD), another autoantigen in type 1 diabetes. This dominant IA-2 epitope peptide also has 75-45% identity and 88-64% similarity over 8-14 aa to sequences in Dengue, cytomegalovirus, measles, hepatitis C, and canine distemper viruses, and the bacterium Haemophilus influenzae. Three other IA-2 epitope peptides are 71-100% similar over 7-12 aa to herpes, rhino-, hanta- and flaviviruses. Two others are 80-82% similar over 10-11 aa to sequences in milk, wheat, and bean proteins. Further studies should now be carried out to directly test the hypothesis that T-cell activation by rotavirus and possibly other viruses, and dietary proteins, could trigger or exacerbate beta-cell autoimmunity through molecular mimicry with IA-2 and (for rotavirus) GAD.

Dendritic cells generated from human blood in granulocyte macrophage-colony stimulating factor and interleukin-7.

Dendritic cells (DC), with potentially important clinical applications, have been generated from human peripheral blood monocytes in the presence of GM-CSF and IL-4 (G4 DC). In the present report we show that DC with a novel phenotype can be generated from blood adherent mononuclear cells in the presence of GM-CSF and IL-7 (G7 DC). Adherent cells from PBMC, cultured in GM-CSF (600 U/ml) and IL-7 (6 U/ml), were transformed over 7 days into cells with DC morphology, at a yield of 1.2-1.6 x 10(6) per 10(7) PBMC. G7 DC not only expressed class I and class II MHC, CD1a, CD11c, CD23, CD40, CD54, CD58, CD80, CD86 and CD95, like G4 DC, but also CD21, which is the complement receptor type 2, a ligand for CD23 and a receptor for EBV and IFN-alpha. G7 DC were at least one log more effective in the autologous MLR and at least two logs more effective in the allogeneic MLR, than PBMC. They elicited proliferative responses of CD4 T cells to tetanus toxoid and CD8 T cells to an EBV peptide, and stronger T-cell cytotoxicity to EBV peptide than G4 DC. Expression of CD21 by G7 DC suggests that IL-7 delivers a distinct signal to DC precursors and that G7 DC may be functionally distinct.

High T cell responses to the glutamic acid decarboxylase (GAD) isoform 67 reflect a hyperimmune state that precedes the onset of insulin-dependent diabetes.

Pancreatic islet beta-cell destruction leading to insulin-dependent diabetes mellitus (IDDM) is an autoimmune T cell-mediated process. Peripheral blood T cells, which proliferate to islet antigens such as glutamic acid decarboxylase (GAD), (pro)insulin or tyrosine phosphatase IA-2, can be detected in at-risk, first degree relatives of people with IDDM. However, cross-sectional studies cannot define the relationship between T cell responses and progression to IDDM. Longitudinal studies were therefore undertaken on 50 at-risk, first degree relatives tested at least yearly for up to 4 years, during which time five developed IDDM. Peripheral blood T cell responses to a GAD67(aa208-404)-glutathione-S-transferase (GST) fusion protein, GST, insulin and tetanus toxoid were measured, together with antibodies to islet cells, GAD, insulin and IA-2. High levels of antibodies to GAD or insulin were generally associated with low T cell responses to these antigens. Relatives who developed IDDM were characterized by high levels of antibodies to insulin and/or islet cells, and high T cell responses to GAD67-GST and tetanus, but not insulin, in the 24 months before clinical diagnosis. Cross-sectionally, T cell responses to GAD67(aa208-404)-GST and to full-length GAD65-GST were highly correlated (r=0.75, P<0.002). In conclusion, increased cellular immunity to the mid region of GAD67 was a marker of late pre-clinical IDDM, but appears to reflect a more general, transient state of cellular immune hyperresponsiveness.

A peptide-binding motif for I-A(g7), the class II major histocompatibility complex (MHC) molecule of NOD and Biozzi AB/H mice.

The class II major histocompatibility complex molecule I-A(g7) is strongly linked to the development of spontaneous insulin-dependent diabetes mellitus (IDDM) in non obese diabetic mice and to the induction of experimental allergic encephalomyelitis in Biozzi AB/H mice. Structurally, it resembles the HLA-DQ molecules associated with human IDDM, in having a non-Asp residue at position 57 in its beta chain. To identify the requirements for peptide binding to I-A(g7) and thereby potentially pathogenic T cell epitopes, we analyzed a known I-A(g7)-restricted T cell epitope, hen egg white lysozyme (HEL) amino acids 9-27. NH2- and COOH-terminal truncations demonstrated that the minimal epitope for activation of the T cell hybridoma 2D12.1 was M12-R21 and the minimum sequence for direct binding to purified I-A(g7) M12-Y20/K13-R21. Alanine (A) scanning revealed two primary anchors for binding at relative positions (p) 6 (L) and 9 (Y) in the HEL epitope. The critical role of both anchors was demonstrated by incorporating L and Y in poly(A) backbones at the same relative positions as in the HEL epitope. Well-tolerated, weakly tolerated, and nontolerated residues were identified by analyzing the binding of peptides containing multiple substitutions at individual positions. Optimally, p6 was a large, hydrophobic residue (L, I, V, M), whereas p9 was aromatic and hydrophobic (Y or F) or positively charged (K, R). Specific residues were not tolerated at these and some other positions. A motif for binding to I-A(g7) deduced from analysis of the model HEL epitope was present in 27/30 (90%) of peptides reported to be I-A(g7)-restricted T cell epitopes or eluted from I-A(g7). Scanning a set of overlapping peptides encompassing human proinsulin revealed the motif in 6/6 good binders (sensitivity = 100%) and 4/13 weak or non-binders (specificity = 70%). This motif should facilitate identification of autoantigenic epitopes relevant to the pathogenesis and immunotherapy of IDDM.

Strategies for identifying and predicting islet autoantigen T-cell epitopes in insulin-dependent diabetes mellitus.

T cells recognize peptide epitopes bound to major histocompatibility complex molecules. Human T-cell epitopes have diagnostic and therapeutic applications in autoimmune diseases. However, their accurate definition within an autoantigen by T-cell bioassay, usually proliferation, involves many costly peptides and a large amount of blood. We have therefore developed a strategy to predict T-cell epitopes and applied it to tyrosine phosphatase IA-2, an autoantigen in IDDM, and HLA-DR4(*0401). First, the binding of synthetic overlapping peptides encompassing IA-2 was measured directly to purified DR4. Secondly, a large amount of HLA-DR4 binding data were analysed by alignment using a genetic algorithm and were used to train an artificial neural network to predict the affinity of binding. This bioinformatic prediction method was then validated experimentally and used to predict DR4 binding peptides in IA-2. The binding set encompassed 85% of experimentally determined T-cell epitopes. Both the experimental and bioinformatic methods had high negative predictive values, 92% and 95%, indicating that this strategy of combining experimental results with computer modelling should lead to a significant reduction in the amount of blood and the number of peptides required to define T-cell epitopes in humans.

Similar peptides from two beta cell autoantigens, proinsulin and glutamic acid decarboxylase, stimulate T cells of individuals at risk for insulin-dependent diabetes.

BACKGROUND: Insulin (1) and glutamic acid decarboxylase (GAD) (2) are both autoantigens in insulin-dependent diabetes mellitus (IDDM), but no molecular mechanism has been proposed for their association. We have identified a 13 amino acid peptide of proinsulin (amino acids 24-36) that bears marked similarity to a peptide of GAD65 (amino acids 506-518) (G. Rudy, unpublished). In order to test the hypothesis that this region of similarity is implicated in the pathogenesis of IDDM, we assayed T cell reactivity to these two peptides in subjects at risk for IDDM. MATERIALS AND METHODS: Subjects at risk for IDDM were islet cell antibody (ICA)-positive, first degree relatives of people with insulin-dependent diabetes. Peripheral blood mononuclear cells from 10 pairs of at-risk and HLA-DR matched control subjects were tested in an in vitro proliferation assay. RESULTS: Reactivity to both proinsulin and GAD peptides was significantly greater among at-risk subjects than controls (proinsulin; p < 0.008; GAD; p < 0.018). In contrast to reactivity to the GAD peptide, reactivity to the proinsulin peptide was almost entirely confined to the at-risk subjects. CONCLUSIONS: This is the first demonstration of T cell reactivity to a proinsulin-specific peptide. In addition, it is the first example of reactivity to a minimal peptide region shared between two human autoimmune disease-associated self antigens. Mimicry between these similar peptides may provide a molecular basis for the conjoint autoantigenicity of proinsulin and GAD in IDDM.

Analysis of families at risk for insulin-dependent diabetes mellitus reveals that HLA antigens influence progression to clinical disease.

BACKGROUND: Individuals at risk for insulin-dependent diabetes mellitus (IDDM), with an affected first-degree relative, can be identified by the presence of islet cell antibodies (ICA). ICA-positive relatives progress at variable rates to IDDM and identification of those at highest risk is a prerequisite for possible preventative treatment. Those who develop IDDM may exhibit less genetic heterogeneity than their ICA-positive or ICA-negative relatives. Specific human leucocyte antigen (HLA) genes predispose to IDDM but could also influence the rate of progression of preclinical disease. Therefore, by comparing HLA antigen frequencies between first-degree relatives, we sought to identify HLA genes that influence progression to IDDM. MATERIALS AND METHODS: HLA antigen frequencies were compared in 68 IDDM, 53 ICA-positive, and 96 ICA-negative first-degree relatives from 40 Caucasoid families. Predictions were tested in a panel of 270 unrelated IDDM subjects. HLA typing was performed serologically (HLA class I and II) and by sequence-specific oligotyping (11th International Histocompatibility Workshop protocol) (HLA class II). ICA tests were measured by an internationally standardized indirect immunofluorescence assay. RESULTS: In general, known susceptibility class II HLA antigens increased in frequency and known protective class II HLA antigens decreased in frequency, from ICA-negative to ICA-positive to IDDM relatives. Thus, DR4 and DQ8 increased whereas DR2 and DQ6 decreased; DR3 and DQ2 increased from ICA-negative to ICA-positive relatives, but not further in IDDM relatives. The high-risk DR3, 4 phenotype increased across the three groups; DR4,X was unchanged, and DR3,X and DRX,X both decreased. The HLA class I antigen, A24, occurred more frequently in ICA-positive relatives who developed IDDM and, in 270 unrelated IDDM subjects, was associated with an earlier age at diagnosis of IDDM in those with the lower risk class II phenotypes DR4,4 and DR3,X. CONCLUSIONS: HLA-DR3 and DQ2 predispose to islet autoimmunity, but not development of clinical IDDM in the absence and DR4 and DQ8. DR4 and DQ8 predispose to the development of clinical IDDM in ICA-positive relatives, in combination with DR3-DQ2 and other haplotypes but not when homozygous. HLA-A24 is significantly associated with rapid progression to IDDM in ICA-positive relatives and with an earlier age at clinical diagnosis. Analysis of IDDM families reveals that HLA genes not only predispose to islet autoimmunity but influence progression to clinical disease. The findings have implications for identifying high-risk relatives as candidates for possible preventative therapy.