Assessment of CD4+ T Cell Responses to Glutamic Acid Decarboxylase 65 Using DQ8 Tetramers Reveals a Pathogenic Role of GAD65 121–140 and GAD65 250–266 in T1D Development

Susceptibility to type 1 diabetes (T1D) is strongly associated with MHC class II molecules, particularly HLA-DQ8 (DQ8: DQA1*03:01/DQB1*03:02). Monitoring T1D-specific T cell responses to DQ8-restricted epitopes may be key to understanding the immunopathology of the disease. In this study, we examined DQ8-restricted T cell responses to glutamic acid decarboxylase 65 (GAD65) using DQ8 tetramers. We demonstrated that GAD65121–140 and GAD65250–266 elicited responses from DQ8+ subjects. Circulating CD4+ T cells specific for these epitopes were detected significantly more often in T1D patients than in healthy individuals after in vitro expansion. T cell clones specific for GAD65121–140 and GAD65250–266 carried a Th1-dominant phenotype, with some of the GAD65121–140-specific T cell clones producing IL-17. GAD65250–266-specific CD4+ T cells could also be detected by direct ex vivo staining. Analysis of unmanipulated peripheral blood mononuclear cells (PBMCs) revealed that GAD65250–266-specific T cells could be found in both healthy and diabetic individuals but the frequencies of specific T cells were higher in subjects with type 1 diabetes. Taken together, our results suggest a proinflammatory role for T cells specific for DQ8-restricted GAD65121–140 and GAD65250–266 epitopes and implicate their possible contribution to the progression of T1D.


Introduction
Type 1 diabetes (T1D) results from destruction of the insulinproducing beta cells of the pancreas. A number of genes have been implicated in T1D development, but genes within the HLA class II region confer most of the disease risk [1]; in particular, it is estimated that 90% of T1D subjects have either an HLA-DQ8 or DQ2 allele [2]. Subjects with the DRB1*04:01 (DR0401)-DQ8 haplotype have an odds ratio of 8.4 for T1D [3], and the predisposing effect of this haplotype is supported by meta-analysis of data sets from different geographic regions [4]. In accordance with the importance of MHC class II molecules in antigen presentation, DQ8-restricted CD4+ T cells are likely to have an essential and pathogenic role in the progression of T1D.
A special feature of DQ8 is the absence of an aspartic acid residue at position 57 of the beta chain [5]. Lack of this Asp residue at beta 57 leads to reduced affinity for antigenic peptides, giving rise to diabetic pathology as a result of ineffective tolerance induction in the thymus [6,7]. Identification and characterization of DQ8-restricted self-epitopes may be key to comprehending DQ8-mediated autoimmunity. A number of DQ8-restriced selfepitopes have been identified using DQ8 transgenic mice (for a recent review see [8]). Some of the reported peptides also elicit responses in HLA-DQ8+ individuals. However, monitoring of specific CD4+ T cells during the progression of diabetes, especially direct assessment of responses in the periphery, has been hampered by low frequencies of effector cells [9] and heterogeneity of the disease.
Autoantibodies against insulin (IA), glutamic acid decarboxylase 65 (GAD65), islet tyrosine phosphatase (IA-2), and zinc transporter 8 (ZnT8) have been used as predictive markers for T1D [10][11][12][13]. The appearance of autoantibodies is clear indication of beta cell autoimmunity and the combined measurement of insulin, GAD65, IA-2, and ZnT8 autoantibodies raise the detection rate to 98% at disease onset [12]. Despite their predictive value, some people with autoantibodies never develop diabetes [14]. In addition, the presence of autoantibodies does not necessarily indicate insulitis, the histopathologic hallmark of T1D that is mainly mediated by Tlymphocytes [15,16]. Therefore, identification of T cell biomarkers correlated with pathogenesis of T1D, together with the presence of autoantibodies, will facilitate disease prediction and prevention.
In this study, we investigated DQ8-restricted CD4+ T cell responses to GAD65. Responses of CD4+ T lymphocytes from T1D subjects to GAD65 121-140 and GAD65 250-266 were visualized using DQ8 tetramers after in vitro expansion of antigen-specific cells with the relevant peptides. In vitro responses and intracellular cytokine staining suggested a strong association of GAD65 121-140 and GAD65 250-266 to the disease. Direct ex vivo staining of peripheral blood mononuclear cells (PBMCs) for GAD65 250-266 revealed higher frequencies of CD4+ and CD4+CD45RO+ T cells in the T1D group. Collectively, our work indicates that T cells specific for GAD65 121-140 and GAD65 250-266 may contribute to the pathogenesis of T1D, and suggests GAD65 250-266 -specific T cells as a potential biomarker for T1D.

CD4+
T cell responses to GAD65 121-140 and GAD65 250-266 are detected more often in subjects with type 1 diabetes after in vitro expansion Seven peptides derived from GAD65 were selected in this study to evaluate their relevance to the progression of autoimmune diabetes. All seven peptides are immunogenic in DQ8 transgenic mice [17][18][19], but DQ8-restricted responses in human subjects were not well demonstrated. Among the seven peptides, GAD65 121-140 , GAD65 206-220, GAD65 250-266 bound to DQ8 with strong affinities ( The three strong DQ8 binders were used to stimulate CD4+ T cells from DQ8+ T1D subjects. After two-week stimulation, GAD65 121-140 and GAD65 250-266 elicited CD4+ T cell responses in multiple subjects. Examples of positive tetramer staining for the two epitopes that elicited responses are shown in Fig. 1. Responses to GAD65 206-220 were negative in five T1D subjects examined and were not further investigated. The association of GAD65 121-140 and GAD65 250-266 with T1D was next examined by comparing the prevalence of specific CD4+ T cell responses in T1D patients and healthy controls. GAD65 121-140 -and GAD65 250-266 -specific CD4+ T cell responses were detected more often in T1D subjects than controls (GAD65 121-140 : 6/10 T1D and 0/9 controls (p = 0.0108); GAD65 250-266 : 9/17 T1D and 1/ 11 controls (Table 2, p = 0.0407); p value evaluated by two-tailed Fisher's exact tests), demonstrating a clear association of DQ8restricted GAD65 T cell responses with disease. DQ8-restricted self-epitopes elicit pathogenic responses Self-peptides can modulate autoimmunity by induction of pathogenic (Th1/Th17) or suppressive (Th2/Treg) responses. To examine functional properties of DQ8-restricted self epitopes, GAD65 121-140 -and GAD65 250-266 -specific CD4+ T cells were cloned from in vitro expanded cultures of multiple subjects with type 1 diabetes. Representative tetramer staining for a GAD65 121-140 and a GAD65 250-266 clone is shown in Fig. 2A. These DQ8 tetramer-sorted clones were antigen-specific and the responses could be blocked by anti-DQ (SPVL3) antibodies (Fig. 2B). The GAD65-specific cells were assayed for cytokine production by intracellular cytokine staining (ICS) (Fig. S1A). All GAD65 250-266specific clones produced Th1-type cytokine IFN-c (Fig. 2C). One GAD65 250-266 -specific clone also produced IL-4, but production of IFN-c was still dominant in these cells (Fig. S1B). IL-17 was not detected in GAD65 250-266 cells. The ICS data were in agreement with the cytokine ELISA results as IFN-c was the main cytokine secreted from GAD65 250-266 clones after peptide specific stimulation (Fig. 2D). Conversely, the cytokine profile for the four GAD65 121-140 -specific T clones was more diversified. Two GAD65 121-140 -specific clones isolated from different patients produced IL-17 (Fig. 2C). One of the IL-17-producing cells exhibited a Th1/Th17 double positive phenotype by co-expressing IFN-c. Production of IL-17 was not a result of in vitro priming of naïve cells, as cells isolated from the memory population (CD45RA-CD4+) maintained the Th17-like phenotype (data not shown). The two IL-17-negative GAD65 121-140 -specific clones were positive for IFN-c, IL-4, and IL-10. However, these two clones might have distinct functionality as suggested by the ratio of IFN-c to IL-4 producing cells (Fig. S1B) and the cytokine ELISA results (Fig. 2D). Although multiple clones isolated from one donor might be the same T cell captured at different stages of differentiation, our T cell clone data from different patients suggest that DQ8-restricted GAD65-specific epitopes could induce proinflammatory responses in T1D subjects.

Higher frequencies of GAD65 250-266 -specific T cells are observed in subjects with type 1 diabetes
In vitro responses can be affected by numbers and activity of monocytes and lymphocytes in the blood samples. To confirm the disease association of DQ8/GAD65-specific cells in an unmanipulated state, ex vivo tetramer staining was performed for DQ8specific T cells in DQ8+ patients and DQ8+ controls (average age for patients: 28.8, and for controls: 34.6, p = 0.2809). Direct ex vivo analysis was able to detect GAD65 250-266 -specific CD4+ T cells in DQ8+ subjects, but not for GAD65 121-140 -specific cells  Fig. S2A and S2B). The unsuccessful detection might be due to low avidity or low frequencies of T cells specific for DQ8/ GAD65 121-140 . A representative ex vivo staining for DQ8/ GAD65 250-266 from one T1D and one healthy subject is shown in Fig. 3A. The ex vivo staining for DQ8/GAD65 250-266 is specific as these cells can be co-stained by DQ8/GAD65 250-266 tetramers labeled with two different fluorochromes (PE or APC) (Fig. 3B).
Results of the ex vivo experiments are summarized in Table 3. The average frequencies of GAD65 250-266 -specific cells and GAD65 250-266 -specific CD45RO+ cells were found to be higher in the T1D group compared to the healthy control (Fig. 3C). There was no correlation between frequencies of GAD65 250-266specific CD45RO+ cells and disease duration (r 2 = 0.0218).

Discussion
DQ8 possesses a predisposing non-aspartic acid residue at beta 57 and confers the highest risk to T1D. Substantial efforts have been made to identify DQ8-restricted responses which can closely monitor the progression of diabetes in human. However, detection of disease-related reactions from the rare islet-specific T cell populations is extremely challenging. In this study, we demonstrated DQ8 tetramers as a tool to study islet-specific T cell reactivity and identified two disease-associated peptides, GAD65 121-140 and GAD65 250-266 (Table 2, Fig. 2). In addition, T cells specific for GAD65 250-266 possessed several attributes that render these cells a potential biomarker for T1D. GAD65 250-266 - Figure 1. Detection of DQ8-specific self-reactive T cells in T1D patients. CD4+ T cells from T1D patients were stimulated with self-peptides derived from GAD65. The cells were cultured for two weeks and stained with PE-tetramers. CD4+ T cell responses to GAD65 121-140 and GAD65 250-266 were detected in multiple subjects. Representative tetramer staining of DQ8-specific self-reactive T cells was shown. Left column: negative staining from a healthy control. Middle and right columns: positive staining from two different T1D patients. Subjects were considered tetramer-positive if a distinct population that was more than three-fold above background in the same experiment was labeled. doi:10.1371/journal.pone.0112882.g001 specific cells exhibited robust responses upon peptide stimulation (Fig. 1), and could be detected directly in unmanipulated PBMCs (Fig. 3A). Most importantly, differences between patients and healthy subjects in frequencies of CD4+ and CD4+CD45RO+ cells specific for GAD65 250-266 could readily be seen by ex vivo analysis (Fig. 3C). Monitoring of GAD65 250-266 -specific T cells can potentially provide new insights into the etiology and treatment of the disease, although further longitudinal studies to link specific responses to the beta cell function are essential to confirm its suitability as a biomarker. Schneider et al. suggest that self-reactive T effector cells are less resistant to peripheral regulation in T1D patients [20]. The current GAD65 121-140 -and GAD65 250-266 -reactive T cell data support this model as responses to both peptides were more prevalent in T1D patients compared to controls (Fig. 1). Higher numbers of total antigen specific cells and CD4+CD45RO+ cells for GAD65 250-266 (Fig. 3C) were also found in the periphery of subjects with T1D by ex vivo tetramer staining. Other studies show that DRB1*04:01 (DR0401)-restricted GAD65-reactive T cells of certain specificities are preferentially activated in T1D patients while other specificities are not [21][22][23][24]. The discrepancy in GAD65 responses to peripheral regulation might reflect that self-reactive T cells for different specificities are subjected to various degree of regulation. Further study in examining whether discrepancy in regulation of DR0401 and DQ8 restricted responses will be of interest. Previous studies show some isletreactive T cells exhibit protective phenotype in healthy subjects [25,26]. The presence of multiple effector properties for GAD65 121-140 (Fig. 2C, Fig. 2D, and Fig. S1B) implies the potential of GAD65 121-140 -specific cells to modulate the immune balance between tolerance and autoimmunity. Future studies to compare the dynamics of T cell responses for each HLA haplotype during disease development will help to shed more insight into the immunopathology of T1D.
The presence of self-reactive T cells in healthy individuals highlights the importance of peripheral regulation as a checkpoint in the progression of T1D. The higher frequencies of GAD65 250-266 -specific total and memory CD4+ T cells in subjects with type 1 diabetes (Fig. 3C) suggest that quantity and phenotypic changes for self-reactive T cells might be an indicator for loss of tolerance. DQ8 tetramers can therefore be a useful tool for the identification of disease-related self-reactive T cells as reliable biomarkers. Monitoring of these cells could assist with early-detection of risk or progression to T1D. In conclusion, DQ8-restricted CD4+ T cells specific for GAD65 121-140 and GAD65 250-266 were detected in multiple T1D subjects. GAD65 121-140 -and GAD65 250-266 -reactive T cells were more prevalent in T1D patients. Further evaluation of T cell responses for beta cell autoantigens should facilitate our understanding of the role of autoreactive DQ8 T cells in type 1 diabetes and selection of autoreactive DQ8 T cells as T1D biomarkers.

Subjects
The studies were approved by the Institutional Review Board of Benaroya Research Institute (BRI, Seattle, WA). All DQ8+ subjects were volunteers of Caucasian descent and were recruited with written consent from participating individuals or their guardians. HLA typing was conducted by BRI sequencing and genotyping core facilities. Samples from DQ8-positive T1D (n = 33, age from 9 to 56, days of disease duration from 281 to 7507) or DQ8-positive individuals without T1D or other autoimmune disease (healthy controls, n = 23, age from 18 to 56) were obtained under the auspices of the BRI JDRF Center for Translational Research at Benaroya Research Institute and Seattle Children's.

DQ8 protein and tetramers
Recombinant DQ8 protein was produced as previously described [27]. Briefly, soluble DQ8 was purified from insect cell culture supernatants by affinity chromatography and dialyzed against citric/phosphate storage buffer, pH 5.4. For the preparation of HLA class II tetramers, DQ8 protein was in vivo biotinylated in Drosophila S2 cells [28] prior to harvest and dialyzed into citric/phosphate buffer. The biotinylated monomer was loaded with 0.2 mg/ml of peptide by incubating at 37uC for 72 h in the presence of 0.2 mg/ml n-Dodecyl-b-maltoside and 1 mM Pefabloc SC (Sigma-Aldrich, St. Louis, MO). Peptide loaded monomers were subsequently conjugated into tetramers using R-PE streptavidin (Biosource International, Camarillo, CA) at a molar ratio of 8:1.

Peptides
Peptides derived from GAD65 (Table 1) were synthesized by Mimotopes (Clayton, Australia). The biotinylated reference peptide GAD65 250-266 (AMMIARFKMFPEVKEKG) was synthesized by Genscript with one 6-aminohexanoic acid spacer added between the N-terminal biotin label and the remainder of the peptide sequence.

Peptide binding competition
GAD65 250-266 is a good binder for DQ8 [29] and was used as the index peptide in the competition study. Various concentrations of each test peptide were incubated in competition with 0.04 mM biotinylated GAD65 250-266 peptide in wells coated with DQ8 protein. After washing, the remaining biotin-GAD65 250-266 peptide was labeled using europium-conjugated streptavidin (Perkin Elmer) and quantified using a Victor2 D time resolved fluorometer (Perkin Elmer). Peptide binding curves were simulated by non-linear regression with Prism software (Version 5, GraphPad Software Inc.) using a sigmoidal dose-response curve. IC 50 binding values were calculated from the resulting curves as Figure 2. Functional analysis of DQ8-specific self-reactive T cells. T cell clones isolated by DQ8 tetramers were assayed for specificity and functionality. (A) Tetramer staining for GAD65 121-140 -specific (clone T1D05-C2), GAD65 250-266 -specific (clone T1D04-C1) and unrelated T cell clones. (B) Representative proliferation results of one GAD65 121-140 -and two GAD65 250-266 -specific T cell clones using APCs from a DR0401/DQ8 homozygous individual. Cells were stimulated with specific or irrelevant control peptide in the absence or presence of 20 mg/ml of L243 (HLA-DR blocking antibody) or SPVL3 (HLA-DQ blocking antibody). SI: stimulation index; cpm of specific peptide divided by cpm of irrelevant peptide. (C) Summary of cytokine-positive clones (for definition see Materials and methods) for GAD65 121-140 and GAD65 250-266 . T cell clones were stimulated with 50 ng/mL phorbol 12-myristate 13-acetate and 1 mg/mL ionomycin in the presence of 10 mg/mL Brefeldin A in 1 mL of T cell medium for 4 hours at 37uC. Cells were fixed, permeabilized, stained with antibodies for IFN-c, IL-10, IL-4, and IL-17, and analyzed on a LSRII multicolor flow cytometer. (D) Secretion of IFN-c, IL-4, and IL-10 from two GAD65 121-140 -specific and two GAD65 250-266 -specific T cell clones. Clones were stimulated in the presence of 10 mg/ml of specific peptide with DQ8 antigen presenting cells. doi:10.1371/journal.pone.0112882.g002 DQ8-Restricted CD4+ T Cell Responses to Glutamic Acid Decarboxylase 65 PLOS ONE | www.plosone.org the peptide concentration needed for 50% inhibition of reference peptide binding.

Self-reactive T cell identification
PBMCs were prepared from the blood of T1D or healthy DQ8 subjects by Ficoll underlay. CD4+ T cells were isolated using the Miltenyi CD4+ T cell isolation kit, plated in 48-well plates with APCs, and stimulated with 20 mg/ml peptides from the selected peptide set. The cells were cultured for two weeks in the presence of T cell medium (RPMI-1640 (Gibso) +10% pooled human serum +1% penicillin-streptomycin) and IL-2 (Hemagen Diagnostics, Inc.) at 37uC, stained with PE-tetramers and antibodies of interest, and analyzed by flow cytometry. 20000 total events were collected for each staining. Lymphoid cells were gated based on forward and side scatter profile. The negative threshold for gating for Fig. 1 was set based on the staining of samples with empty tetramers. Our criterion for positivity was distinct staining that labeled a compact tetramer positive CD4+ population, and was more than three-fold above background (cells from an unstimulated well) in the same experiment.

Ex vivo analysis of self-reactive T cells
Ex vivo tetramer staining was performed as previously described [30]. Briefly, 30 million PBMCs in culture medium at a concentration of 150 million/mL were incubated with 50 nM dasatinib (LC Laboratories, dissolved in DMSO) for 10 min at 37uC. The cells were next stained with 20 mg/mL phycoerythrin (PE)-labeled tetramers at room temperature for 120 minutes, followed by antibody staining for 20 minutes at 4uC with CD4 APC (clone RPA-T4, eBioscience), CD45RO FITC (clone UCHC1, eBioscience), and a combination of CD14 PerCP (clone MQP9, BD PharMingen) and CD19 PerCP (clone SJ25C1, BD PharMingen) to exclude B cells and monocytes from the analysis. Cells were washed twice with FACS running buffer (0.5% BSA and 2mM EDTA in PBS) and incubated with anti-PE magnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany) at 4uC for another 20 minutes. The cells were washed again and enriched with a Miltenyi MS magnetic column. Samples were labeled with ViaProbe (BD Biosciences) and analyzed on a BD FACS Calibur flow cytometer. Frequencies were calculated by dividing the number of tetramer positive cells in the bound fraction by the number of total CD4+ T cells in the sample.

Statistical analysis
Statistical analysis was performed with Prism software (Version 5.03, GraphPad Software Inc.). The Mann-Whitney U-test was used to preform two group comparisons of T cell frequency. Fisher's exact test was used to compare the prevalence of T cell responses in T1D and healthy subjects.

T cell clone isolation
T cell clones were generated by staining cultured T cells with tetramers, sorting gated tetramer-positive CD4+ cells using a FACS Aria (at single-cell purity, applying a singlet gate to the lymphocyte population) and expanding in a 96-well plate in the presence of 1.0610 5 irradiated PBMCs and 2 mg/ml phytohemagglutinin (Remel Inc. Lenexa, KS). After expansion of each T cell clone to a single 48 well, clones were stained again with tetramer and also stimulated in parallel with the corresponding peptides (10 mg/mL), adding HLA-DQ-matched irradiated PBMCs as antigen presenting cell and measuring thymidine uptake to verify epitope specificity.