Detection of GAD65-Specific T-Cells by Major Histocompatibility Complex Class II Tetramers in Type 1 Diabetic Patients and At-Risk Subjects

Blood samples from patients with type 1 diabetes (n = 4, 14–25 years of age) were obtained 1–6 months postonset. At-risk subjects were positive for two or more autoantibodies (GAD65, the insulinoma-associated tyrosine phosphatase-like protein IA-2, and insulin autoantibody), except for two subjects who had a single autoantibody. All patients were being treated for diabetes at the Section of Endocrinology, Virginia Mason Medical Center, and the at-risk subjects (aged 8–22 years, n = 6) were participants in the prediabetes screening program at the Virginia Mason Research Center. Healthy blood donors (n = 5) were recruited from the hospital/research center staff. Only subjects positive for DR401 or DR404 were included in our study.

The construction of the expression vectors for generation of the soluble DRA*0101/DRB1*0401 has been described previously (16). Briefly, a site-specific biotinylation sequence was added to the 3′ end of the DRB1*0401 or DR404 leucine zipper cassette, and the chimeric cDNA was subcloned into a Cu-inducible Drosophila expression vector. DR-A and DR-B expression vectors were cotransfected into Schneider S-2 cells, purified, concentrated, and biotinylated. Specific peptide was loaded for 48–72 h, and tetramers were formed by incubating class II molecules with phycoerythrin (PE)-labeled streptavidin.

PBMCs were separated from 15–40 ml heparinized blood by gradient centrifugation (Lymphoprep; Nycomed, Oslo, Norway). Cells were resuspended in RPMI-1640 (Gibco/BRL, Rockville, MD) supplemented with 2 mmol/l l-glutamine, 1 mmol/l sodium pyruvate, 100 μg/ml penicillin/streptomycin, and 15% vol/vol pooled human serum at a density of 5 × 10⁶/ml and cultured in the presence of a GAD65 555-567 (557I; NFIRMVISNPAAT) peptide at a concentration of 10 μg/ml. From some samples a control culture was set up with an myelin base protein (MBP) 84-102 (NPVVHFFKNIVTPRTPPP) peptide that binds well to DR401 and -404. On days 6–10, the cells were transferred at a density of 4 × 10⁶/ml onto a 48-well plate that had been absorbed with 10 μg/ml DR401 or -404 monomer in 1 × PBS for 2 h at 37°C. The class II monomer contained the same peptide used in the primary culture. Then, 1 μg/ml anti-CD28 antibody (BD/Pharmingen, San Jose, CA) was added into the culture, and a low dose of interleukin-2 (IL-2; 2.5 units/ml) was added 48 h later.

On day 3 the stimulated cells were stained by 10 μg/ml of PE-labeled HLA-DR401 or HLA-DR404 tetramer for 3 h at 37°C and subsequently by fluorochrome-labeled anti-CD25 and anti-CD4 (BD/Pharmingen, San Jose, CA) for 30 min on ice. Cells were then washed with PBS containing 1% fetal bovine serum and analyzed using a Becton-Dickinson FACSCalibur flow cytometer. The calibration was performed by using cells stained by single-fluorochrome anti-CD4. Data analysis was performed by using WinMdi (Stanford University) and CellQuest (Becton Dickinson) software programs.

The top 1% of the CD4 staining intensity among the CD25/CD4^high cells from two at-risk subjects were single-cell sorted into 96-well plates using a FACS Vantage cell sorter (Becton Dickinson). Clones were expanded for 10 days by stimulation with irradiated unmatched PBMCs (1.5 × 10⁵/well), 5 μg/ml phytohemagglutinin, and 10 units/ml IL-2. Subsequently, the cells were stimulated by HLA-DR4⁺ PBMCs pulsed with 10 μg/ml of GAD65 557I peptide and 10 units/ml IL-2, and on day 7 they were selected by growth for further expansion. The resting T-cells were tested for proliferation by stimulating 2 × 10⁴ T-cells with irradiated autologous PBMCs (5–10 × 10⁴/well) with and without a specific peptide. ³H-thymidine incorporation was measured at 72 h.

PBMCs from four new-onset type 1 diabetic patients were stimulated with GAD65 557I peptide for 6–10 days. This peptide, NFIRMVISNPAAT, has an F-to-I substitution at position 557, which has been shown to enhance agonist activity for proliferation and cytokine release from DR4-restricted T-cells (20a). After the primary PBMC culture, the cells were stimulated with immobilized DR401 or -404 monomer containing GAD65 557I or irrelevant peptide, and soluble anti-CD28 antibody. On day 3, the cells were stained by fluorochrome-conjugated specific and control tetramers and anti-CD25 and anti-CD4 surface markers and analyzed by flow cytometry. All four patients displayed a highly activated T-cell subset expressing a CD25⁺/CD4^high+ phenotype (Figs. 1A– D). This activation phenotype was not present in normal subjects (Figs. 1E– F).

We investigated GAD65 epitope-specific T-cell activation in six at-risk subjects (Fig. 2) who were positive for either HLA-DRB1*0401 (subject 5574, 7657, 7878, 6827, and 6899) or DRB1*0404 (6212). Four subjects (6212, 5574, 7657, and 7878) were positive for two or more autoantibodies. One subject (6827) was positive for low levels of insulin autoantibodies and had impaired glucose tolerance, and one subject (6899) had only GAD65 autoantibodies. Three of the subjects were followed at two occasions (Figs. 2A– C). Subjects 7878, 6212, and 5574 (Figs. 2A– B and D) have a distinct subset of CD25⁺/CD4^high+-positive T-cells, as was observed in all our patients with type 1 diabetes. The sample from subject 7878 (Fig. 2A) shown on the left panel was drawn 5 months before the subsequent one shown on the right panel. The T-cell activation profile shows a dramatic difference between these two time points. The T-cells from the first sample show very little activation when stimulated with the GAD65 557I peptide, in contrast to the T-cells obtained a few months later, which express a distinct CD25⁺/CD4^high+ phenotype. Whether this difference in the GAD65-specific T-cell response indicates progression to type 1 diabetes remains to be seen. The second at-risk subject, 6212 (Fig. 2B), displayed an activated CD25⁺/CD4^high+ phenotype in the first sample (left panel), and no change was observed 7 months later (right panel). The third at-risk subject, 6827 (Fig. 2C), displays a decrease in the number of CD25⁺ cells in the sampling 5 months later (right panel), but on both occasions the phenotype of these cells was CD4^low. Also, subject 7657 (Fig. 2E) has CD25⁺ T-cells, but they are predominantly the CD4^low phenotype. Interestingly, the two subjects who display the least T-cell activation (Figs. 2C and F) have only single islet cell-specific autoantibody, suggesting a lower risk to develop type 1 diabetes.

Figure 3 shows three examples of tetramer binding analysis, using HLA-DR401 GAD65 tetramers to stain cells gated on CD25⁺/CD4^high+ markers. Figure 3A shows tetramer staining of the sample from type 1 diabetic patient 7810 shown in Fig. 1A. Of the cells expressing CD25⁺/CD4^high+ phenotype, 28.7% stained with the specific GAD65 tetramer. Binding to herpes simplex virus (HSV) p61-control tetramer was 0.8%. Figure 3B shows tetramer staining in at-risk subject 7878. The gating criteria were the same as in Fig. 3A. Of the CD25⁺/CD4^high+ cells, 5.2% bound the GAD65 tetramer. Figure 3C shows the lack of tetramer staining in at-risk subject 7657, who had an increased number of CD25⁺/CD4^low cells (shown in Fig. 2E). Because this subject lacked the subset of CD4^high+ cells, the gating was set on the top 25% of the CD4 staining intensity among CD25⁺/CD4⁺ cells, but no tetramer-positive cells were detected. Whether the activated CD25⁺ cells are indeed antigen specific but have an affinity too low to stain by tetramer remains to be investigated. Overall, these findings indicate that CD4⁺ T-cells that are able to bind GAD tetramers reside in the highly activated antigen-specific cell population characterized by the simultaneous high level of expression of CD25 and CD4.

Table 1 illustrates the relationship between tetramer staining and T-cell activation to GAD65 in our type 1 diabetic, at-risk, and normal subjects. All four new-onset diabetic patients had a highly activated CD25⁺/CD4^high+ cell population that was induced upon stimulation by GAD65 557I peptide, and 4–28% of these cells also had the ability to bind the GAD65 tetramer. The same activation profile including positive tetramer staining was also observed in two at-risk subjects. In one of these subjects, the activated phenotype of CD4⁺ T-cells was not present in an earlier sample, consistent with an emerging T-cell response toward this particular epitope. Heterogeneity in the GAD65-specific T-cell response among at-risk individuals, in terms of the expression of activation markers and tetramer staining that might be associated with the disease progression, is intriguing, but correlation with outcome requires long-term follow-up. Activation of T-cells to the GAD peptide, indicated by either CD25 or CD4^high expression, was not detected in normal subjects. Also, when PBMCs from type 1 diabetic patients or at-risk subjects were stimulated by irrelevant peptide (MBP) in the primary culture and then exposed to control immobilized empty class II MHC, no activation or tetramer staining was observed (data not shown).

Specificity of the T-cell activation profile was examined in a more detailed fashion, using single-cell sorting of the CD25⁺/CD4^high+ cells from two at-risk subjects, 6212 and 5574, using flow cytometry to select cells from the top 1% of the CD4⁺ staining intensity. Seven clones were obtained from both subjects, and these clones proliferated consistently in the presence of GAD65 557I peptide in replicate experiments. The T-cell proliferation in the presence of peptide was dose-dependent, and the clones responded to both the 557-I superagonist and the wild-type peptide GAD65 555-567 (Fig. 4). A clone isolated from subject 5574 stained strongly with GAD65 tetramer (Fig. 5).

The use of soluble class II MHC tetramers has enabled the identification of antigen-specific T-cells in type 1 diabetic patients and some at-risk subjects. Our approach takes advantage of the appearance of highly activated T-cells expressing a CD25⁺/CD4^high+ phenotype induced by immobilized class II MHC monomer containing the GAD65 peptide. This approach has enabled us to demonstrate in type 1 diabetic patients and some at-risk subjects a HLA-DR4 monomer-induced activation profile characterized by upregulation of CD4 on GAD65-specific T-cells. Almost all T-cells that stain with the specific tetramer reside in this population, and because this activation profile is not present in normal subjects, it could provide a useful tool for analysis of the T-cell response in autoimmune diabetes.

CD4⁺ T-cell responses in recent-onset diabetic patients have been previously investigated in vitro using GAD peptides to induce proliferation of PBMCs. A major problem in detecting and quantitating autoreactive T-cells is their extremely low frequency in the peripheral blood of patients. Recent studies of patients with infections or autoimmunity, using class II tetramers, emphasize this point. In one study, the presence of collagen II- or glycoprotein 39-specific CD4⁺ T-cells in synovial fluid of patients with rheumatoid arthritis was investigated (21). Although DR4 tetramers stained antigen-specific hybridomas, no tetramer-positive cells could be detected in synovial fluid of the patients. However, in a small number of DR401⁺ patients, a low level of staining, slightly higher than background, was detected by glycoprotein 39 tetramer. It is possible that the number of T-cells recognizing these epitopes was too low for effective identification using class II tetramers. Also, the study was performed with patients who were 5 years postonset, and it is possible that the T-cell response to these particular epitopes from glycoprotein 39 and collagen II would have been stronger closer to disease onset. Another study demonstrated that class II tetramers can be used to enumerate CD4⁺ T-cells specific for Borrelia epitope outer surface protein A (OspA) from the synovial fluid and peripheral blood of HLA-DR401⁺ patients with treatment-resistant Lyme arthritis (22). Although the frequency of tetramer-positive cells was very low, the authors could generate OspA-specific clones by sorting the tetramer-positive cells. In contrast, studies of CD4⁺ T-cell responses to influenza or HSV 2 antigens have shown efficient detection with class II tetramers after in vitro expansion of responding cells (16,17).

The problems associated with the detection of rare autoreactive T-cells in the periphery by using tetramers has been addressed by Liu et al. (23) in a study on GAD-reactive CD4⁺ T-cells in the NOD mouse model. T-cells from immunized mice were readily detectable with a I-A^g7 tetramer covalently linked to GAD peptides, but those from islets or lymphoid organs of untreated NOD mice were not readily detectable, suggesting that immunization may expand antigen-specific T-cells or cells expressing relatively high-affinity T-cell receptors (TcRs), allowing better staining by tetramer. In our study the activation and in vitro amplification of GAD65-specific T-cells was crucial for detection of tetramer-positive cells. When tetramer staining was performed after a 10-day primary culture without secondary stimulation by GAD65-DR401 monomer, no tetramer-positive cells were detected (data not shown). We did not perform a direct comparison between antigen-presenting cells (APCs) and immobilized monomer stimulation because of the low number of cells available for analysis, but it is very likely that the DR401 monomer system provides a higher-density interaction between MHCs and TcRs. Problems associated with in vitro amplification include potential changes in the phenotype of the CD4⁺ antigen-specific T-cells, preferential proliferation of high-affinity T-cells, and activation-induced cell death. In our T-cell culture system, a GAD65 557I peptide was used for better binding to DR401/404 and for the superagonist nature in the stimulation of T-cell clones specific for this minimal epitope. This high-binding peptide on a immobilized DR401/404 monomer may be one explanation for the efficient induction of the strongly activated CD25⁺/CD4^high+ cell population and expansion of these cells in the GAD65-responsive subjects. Notably, however, these CD25⁺/CD4^high+ T-cells, though expanded by GAD65 557I, also responded to the wild-type peptide GAD65 555-567, demonstrating true autoreactivity.

A correlation between T-cell activation, CD25⁺/CD4^high+ phenotype, and tetramer staining occurred in subjects who have type 1 diabetes or signs of diabetes-associated autoimmunity. Tetramer staining has been shown to be dependent on the activation state of CD4⁺ T-cells in a study by Cameron et al. (24), in which they blocked staining by HA-DR1 oligomers by treatments that interfered with cytoskeletal rearrangements and endocytosis. Restimulation of antigen-specific T-cells with peptides in vitro can upregulate their CD4 expression. Liu et al. (23) demonstrated in their study that almost all the cells bound by I-A^g7-GAD tetramer had upregulated their CD4 expression and become CD4^high+, although a range in tetramer staining intensity was observed. This finding was extended to human systems in a study by Novak et al. (19). The physiological relevance of CD4 upregulation in TcR-MHC interaction is not fully understood, but it may contribute to signal amplification (25) or prolong the interaction between the TcR and MHC by affecting the dissociation of the complex (26).

One of our at-risk subjects had a high number of CD25⁺/CD4⁺ cells with low levels of tetramer binding. It is possible that these cells are antigen-specific but possess low-affinity TcR. It has been demonstrated by Crawford et al. (27) that tetramer binding is directly proportional to TcR affinity and the level of TcR expression. Other alternative explanations are that the T-cells with high-affinity TcRs did not survive in the culture, or the CD25⁺ cells represent regulatory T-cells prevalent in the prediabetic period and are possibly associated with a clinically nonprogressive phenotype. Long-term prospective analysis of these at-risk subjects is needed to address this possibility.

The use of tetramers in the search of autoreactive T-cells is based on the presumption that the peptide on the tetramer corresponds to an immunodominant epitope of interest that is prevalent in the disease process. However, in type 1 diabetes, T-cell responses to several other epitopes have been reported, and it has been suggested that epitope spreading may occur during the progression of the disease (4,8,28,29). In addition, there are differences in the processing of epitopes between individuals and different APCs (30–32). All of these variables could create differences in the selection of autoreactive T-cells and could potentially limit the utility of the approach described here. Nevertheless, the use of tetramer techniques in the detection of autoreactive T-cells in type 1 diabetic patients and at-risk subjects is a powerful tool to gain insight into the mechanisms of molecular basis of autoimmunity. The phenotyping of T-cells should provide useful markers for progression of immune-mediated β-cell reactivity and could be used in clinical trials to evaluate the efficacy of the immunomodulatory therapies targeting intervention/prevention of type 1 diabetes.

The study was supported by grants from the National Institutes of Health and the Juvenile Diabetes Research Foundation. H.R. was supported by Juvenile Diabetes Research Foundation International Career Development Award Grant 2-200-144.

We thank Dr. Vivian Gersuk for HLA-typing, Tuan Nguyen for processing the blood samples, Dr. John Gebe for expert advice in data processing, and Janice Abbas for preparation of the manuscript. We also thank Dr. George Eisenbarth and Ben Falk for critical reading of the manuscript.