Research ArticleAUTOIMMUNITY

Increased islet antigen–specific regulatory and effector CD4+ T cells in healthy individuals with the type 1 diabetes–protective haplotype

See allHide authors and affiliations

Science Immunology  14 Feb 2020:
Vol. 5, Issue 44, eaax8767
DOI: 10.1126/sciimmunol.aax8767

Dominant protection against type 1 diabetes

Polymorphic histocompatibility genes in the HLA locus have a strong influence on genetic susceptibility to type 1 diabetes (T1D). Several high-risk HLA haplotypes increase susceptibility to T1D, whereas the DR1501-DQ6 HLA haplotype confers dominant protection. Wen et al. investigated the mechanistic basis for this protective effect by measuring the frequency of CD4+ T cells reactive with epitopes on islet autoantigens in healthy individuals. Individuals with the protective HLA haplotype had a higher frequency of islet antigen–reactive CD4+ T cells than individuals with high-risk or neutral haplotypes. These CD4+ T cells included both regulatory T cells and effector T cells with a propensity to make IL-10. These findings provide deeper insights into how beneficial subsets of self-reactive T cells maintain prophylaxis against autoimmunity.


The DRB1*15:01-DQB1*06:02 (DR1501-DQ6) haplotype is linked to dominant protection from type 1 diabetes, but the cellular mechanism for this association is unclear. To address this question, we identified multiple DR1501- and DQ6-restricted glutamate decarboxylase 65 (GAD65) and islet-specific glucose-6-phosphatase catalytic subunit–related protein (IGRP)–specific T cell epitopes. Three of the DR1501/DQ6-restricted epitopes identified were previously reported to be restricted by DRB1*04:01/DRB1*03:01/DQB1*03:02. We also used specific class II tetramer reagents to assess T cell frequencies. Our results indicated that GAD65- and IGRP-specific effector and CD25+CD127FOXP3+ regulatory CD4+ T cells were present at higher frequencies in individuals with the protective haplotype than those with susceptible or neutral haplotypes. We further confirmed higher frequencies of islet antigen–specific effector and regulatory CD4+ T cells in DR1501-DQ6 individuals through a CD154/CD137 up-regulation assay. DR1501-restricted effector T cells were capable of producing interferon-γ (IFN-γ) and interleukin-4 (IL-4) but were more likely to produce IL-10 compared with effectors from individuals with susceptible haplotypes. To evaluate their capacity for antigen-specific regulatory activity, we cloned GAD65 and IGRP epitope–specific regulatory T cells. We showed that these regulatory T cells suppressed DR1501-restricted GAD65- and IGRP-specific effectors and DQB1*03:02-restricted GAD65-specific effectors in an antigen-specific fashion. In total, these results suggest that the protective DR1501-DQ6 haplotype confers protection through increased frequencies of islet-specific IL-10–producing T effectors and CD25+CD127FOXP3+ regulatory T cells.


Type 1 diabetes (T1D) is a multifactorial disease in which both genetic and environmental factors contribute to disease development (17). Genes in the human leukocyte antigen (HLA) region account for up to 50% of the genetic basis of T1D (8). In particular, both the DRB1*04:01-DQA1*03:01-DQB1*03:02 (DR0401-DQ8) and DRB1*03:01-DQA1*05:01-DQB1*02:01 (DR0301-DQ2) haplotypes are susceptible haplotypes, whereas the DRB1*15:01-DQA1*01:02-DQB1*06:02 (DR1501-DQ6) haplotype confers dominant protection. Other haplotypes such as DRB1*07:01-DQA1*02:01/DQB1*02 (DR0701-DQ2.2) are generally considered to be neutral (13, 810). For the DR0401-DQ8 haplotype, DQ8 is the most recognized disease-susceptible allele (1113), but the DR0401 allele also contributes to susceptibility (14, 15). For the protective DR1501-DQ6 haplotype, the strong linkage disequilibrium between the DR1501 allele and the DQ6 allele renders it difficult to dissect out the relative roles of DR and DQ in conferring protection. However, studies of a family with an unusual DR1501-containing haplotype implicated DQ6 rather than DR1501 as the dominantly protective allele (16).

The mechanisms by which the disease-susceptible alleles confer risk have been investigated. Disease-associated alleles such as DQ8 and DQ2 have a small noncharged residue at position 57 of the DQB1 chain (located around the pocket 9 region), which shapes the peptide-binding repertories of these DQ molecules (11, 17). Within the thymus, the threshold of the binding affinity of the T cell receptor (TCR) for peptide–major histocompatibility complex (MHC) will determine whether thymocytes are negatively selected within the thymus or develop into mature T cells and enter the circulation (18). In addition, thymocytes with TCRs that have affinity near the threshold for negative selection will enter into the periphery as thymus-derived regulatory T cells (tTregs) (19). It has been proposed that autoreactive thymocytes from hosts with the susceptible HLA alleles are not as efficiently negatively selected and these cells escape into the periphery as autoreactive T cells (20, 21). tTregs and peripherally induced Tregs (pTregs) (22) counter the activity of these effector T cells (Teffs) and limit autoreactive T cell responses. In individuals with high-risk alleles, additional genetic factors combine with environmental triggers, resulting in the loss of the peripheral regulatory mechanisms, ultimately leading to clinical disease (23).

Multiple mechanisms have been proposed to explain the dominant protection afforded by protective HLA alleles in T1D. Studies of the nonobese diabetic (NOD) mouse model suggested that protective MHCs are effective in deleting autoreactive T cells (24, 25). Another mouse model showed that a protective allele could shape the development of intestinal microbiota and prevent insulitis (26). Studies in human individuals have primarily focused on DQ6 and the epitope stealing hypothesis. Various data have demonstrated that the peptide-binding motif for DQ6 is distinct from the susceptible DQ8 molecules (27, 28). DQ6 can compete with DQ8 for identical antigenic peptides, and presentation of these peptides by the protective HLA leads to the production of anti-inflammatory cytokines than inflammatory cytokines (28, 29). Although these results are provocative, the number of studies that have examined the mechanisms by which the DR1501-DQ6 haplotype confers protection in human individuals remains limited.

One area that has not been comprehensively studied is the possible role of DR1501- and DQ6-restricted islet antigen–specific CD4+ T cells in mediating mechanisms of dominant protection. In the present study, we examined islet antigen–specific CD4+ T cells in healthy individuals with the protective DR1501-DQ6 haplotype. To facilitate these assays, we first identified DR1501- and DQ6-restricted T cell epitopes from islet antigens. We then compared the frequencies of islet antigen–specific CD4+ T cells and Tregs in healthy individuals with protective haplotype, susceptible haplotypes, or neutral haplotype. We further evaluated the cytokine profiles of islet antigen–specific Teffs and the suppressive capacity of DR1501-restricted Tregs, allowing us to evaluate multiple pathways that might contribute to the dominant protection provided via the DR1501-DQ6 haplotype.


Increased frequencies of DR1501-restricted GAD65- and IGRP-specific CD4+ T cells in individuals with the protective haplotype

Glutamate decarboxylase 65 (GAD65) and islet-specific glucose-6-phosphatase catalytic subunit–related protein (IGRP) are important T1D-associated autoantigens (30, 31). Previous studies have demonstrated the presence of GAD65- and IGRP-specific autoreactive CD4+ T cells not only in individuals with T1D but also in healthy individuals with DR0401 or DR0301 haplotypes (3234). In this study, we hypothesized that DR1501- and DQ6-restricted CD4+ T cells are likewise present in healthy individuals with the dominant protective haplotype and that these cells have trait that can be measured to reveal an active role in protection. To facilitate our studies, we applied a previously described tetramer-guided epitope mapping (TGEM) approach to identify DR1501- and DQ6-restricted epitopes within GAD65 and IGRP (34, 35). Through screening experiments with six different DR1501-DQ6–positive healthy individuals, we identified a total of 11 epitopes: 3 DR1501-restricted GAD65, 4 DR1501-restricted IGRP, 3 DQ6-restricted GAD65, and 1 DQ6-restricted IGRP epitopes. The sequences of all these epitopes are summarized in Table 1. Representative results from a TGEM experiment in identifying DR1501-restricted GAD65 epitopes are shown in fig. S1. Specific tetramer staining results for DR1501- and DQ6-restricted GAD65 and IGRP epitope–specific T cells are shown in fig. S2. Responses against all of the DR1501-restricted GAD65 and IGRP T cells were detected in more than 50% of the individuals, suggesting that these epitopes were the most immunodominant. A similar approach was applied to identify DR0701-restricted GAD65 and IGRP epitopes. With the exception of the DR0301-restricted GAD65489–508 epitopes, the DR0401- and DR0301-restricted GAD65 and IGRP epitopes used for assembling tetramers in the current study have been published previously (33, 34, 36). The sequences of these DR0401-, DR0301-, and DR0701-restricted epitopes are also included in Table 1, and representative examples of tetramer staining for positive epitope-specific cell lines are shown in fig. S2.

Table 1 T cell epitopes and tetramers for ex vivo staining.

View this table:

Specific tetramers were then used to examine the frequencies of GAD65- and IGRP-specific CD4+ T cells directly ex vivo in peripheral blood mononuclear cells (PBMCs) of healthy individuals with either a protective haplotype (DR1501-DQ6), susceptible haplotypes (DR0401-DQ8 or DR0301-DQ2), or a neutral haplotype (DR0701-DQ2.2). For each staining, relevant tetramers corresponding to GAD65 or IGRP epitopes were examined separately. Influenza (Flu) tetramers were also included as a positive control. Examples of these staining experiments are shown in Fig. 1A, and the results for all individuals are summarized in Fig. 1B.

Fig. 1 Analysis of GAD65- and IGRP-specific CD4+ T cells by direct ex vivo tetramer staining.

(A) Representative MHC class II tetramer staining of GAD65-, IGRP-, and Flu-specific CD4 T cells in healthy donors. Frequencies of tetramer-positive T cells per million are as indicated. (B) Comparison of the frequency of GAD65- and IGRP-specific CD4+ T cells (left) and CD45RA memory CD4+ T cells (right) among healthy donors. The DR0701-restricted tetramer stainings were performed in individuals with DR0701-DQ2.2 (n = 7). The DR0301-restricted tetramer stainings were performed in individuals with DR0301-DQ2 (n = 6). The DR0401-restricted tetramer stainings were performed in individuals with DR0401-DQ8 (n = 10). The DR1501- and DQ6-restricted tetramer stainings were performed in individuals with DR1501-DQ6 (DR1501, n = 14; DQ6, n = 6). Shown is the mean ± SEM. Welch’s t test was used. N.S., P ≥ 0.05; **P < 0.01; ***P < 0.001; and ****P < 0.0001.

Of particular significance, DR1501-restricted GAD65 T cells had the highest observed frequencies overall. Furthermore, frequencies of DR1501-restricted IGRP-specific CD4+ T cells were higher in DR1501 individuals as compared with IGRP-specific cells in individuals with either high-risk alleles or a neutral allele (Fig. 1, A and B). We also discovered that the GAD65137–156 epitope was the predominant DR1501-restricted epitope because more than 80% of GAD65-specific T cells detected ex vivo exhibited this particular specificity (fig. S3). Additional experiments also showed that GAD65137–156 was naturally processed and presented, because GAD65137–156-specific T cells expanded and could be observed by tetramer staining upon stimulation of PBMCs with GAD65 protein (fig. S4).

High frequencies of GAD65- and IGRP-specific Tregs and Teffs in individuals with the T1D-protective haplotype

CD4+ Tregs are known to be critical for maintaining T cell tolerance and controlling autoimmunity. Multiple studies have shown that human individuals with T1D exhibit defects in the regulatory function of their Tregs (37, 38). Studies in different murine models have shown that boosting Treg number or activity through different approaches can delay or prevent T1D onset (39, 40). Therefore, clinical trials using polyclonal Tregs to treat T1D individuals are currently being pursued (41). Because individuals with protective HLA alleles exhibited relatively high frequencies of islet antigen–specific CD4+ T cells and yet such individuals are not prone to develop T1D, we hypothesized that islet antigen–specific Tregs must be present to mediate protection in these individuals. To address this question, we implemented a staining panel to enumerate Tregs. CD4+CD127CD25+tetramer+ cells were designated as islet antigen–specific Tregs, because cells with this combination of surface markers have been shown to be highly enriched for forkhead box P3 dependent–positive (FOXP3+) T cells (Fig. 2) (42, 43). Frequencies of CD4+CD127+CD25tetramer+ Teffs were also examined, allowing us to evaluate Treg/Teff ratios. As expected, higher frequencies of DR1501-restricted GAD65/IGRP-specific Tregs were observed for individuals in the T1D-protective group than for those in T1D high-risk groups and the neutral-risk DR0701 group (Fig. 3A). We also observed higher frequencies of DR1501-restricted islet antigen–specific Teffs compared with the other groups (Fig. 3B). There were no significant differences observed in the Treg/Teff ratios between groups with the exception of GAD65 DR0701 versus GAD65 DR1501 and IGRP DR0301 versus IGRP DR1501 (Fig. 3C). Overall, our results suggest that higher frequencies of islet antigen–specific Tregs and Teffs are present in individuals with the protective DR1501-DQ6 haplotype.

Fig. 2 Direct ex vivo tetramer staining in conjunction with intracellular staining of FOXP3.

(Top) Representative tetramer staining of GAD65- and IGRP-specific CD4+ T cells in a DR1501-DQ6 healthy donor in combination with intracellular staining. (Bottom) The left and right panels show the intracellular FOXP3 staining of Teff (CD127+CD25) and Treg (CD127CD25+) populations within GAD65-specific (light blue) and IGRP-specific (red) CD4+ T cells and CD45RACD4+ T cells, respectively.

Fig. 3 Analysis of Tregs among T1D high-risk, neutral, or protective groups.

(A) Frequency of GAD65/IGRP antigen–specific CD4+ Tregs (left) and CD45RACD4+ memory Tregs (right) in individuals with T1D high-risk, neutral, or protective HLA haplotypes. (B) Frequency of GAD65/IGRP antigen–specific CD4+ Teffs (left) and CD45RA memory Teffs (right) in individuals with T1D high-risk, neutral, or protective HLA haplotypes. (C) Ratio of GAD65 and IGRP antigen–specific Tregs/Teffs. The DR0301-restricted tetramer stainings were performed in individuals with DR0301-DQ2 (n = 6). The DR0401-restricted tetramer stainings were performed in individuals with DR0401-DQ8 (n = 10). The DR0701-restricted tetramer stainings were performed in individuals with DR0701-DQ2.2 (n = 7). The DR1501- and DQ6-restricted tetramer stainings were performed in individuals with DR1501-DQ6 (n = 14 and n = 6, respectively). Shown is the mean ± SEM. Welch’s t test was used. N.S., P ≥ 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; and ****P < 0.0001.

HLA restriction influences the cytokine profiles of islet antigen–specific CD4+ T cells

To characterize possible differences in the cytokine profiles of islet antigen–specific T cells, we single cell–sorted and expanded DR1501-restricted GAD65/IGRP tetramer+CD45RACD127CD25+ Tregs and CD45RACD127+CD25 Teffs as T cell clones. The specificity of these clones was validated by tetramer staining. Representative of these efforts, multiple DR1501-restricted GAD65137–156 and IGRP225–244-specific Treg and Teff clones were isolated using this strategy. As expected, tetramer+ Treg clones expressed high levels of FOXP3 and Helios, whereas tetramer+ Teff clones did not (Fig. 4, A and B). These data further support the fact that CD45RACD127CD25+ cells were Tregs (4446).

Fig. 4 Single-cell cloning of DR1501-restricted GAD65- or IGRP-specific CD4+CD45RA T cells from individuals with T1D-protective DR1501-DQ6 haplotype.

(A) Individual cells identified by surface staining with DR1501-restricted GAD65 or IGRP tetramers were cloned by single-cell sorting and expanded in vitro and further validated by tetramer staining. Intracellular staining of FOXP3 and Helios was performed on the clones. Representative flow cytometry profiles are shown. (B) Comparison of levels of FOXP3+Helios+ % between 30 Teff clones (specificities of these clones are shown in table S1) and 16 Treg clones (specificities of these clones are shown in table S2) generated from at least six independent experiments. Shown is the mean ± SEM. Welch’s t test was used. ****P < 0.0001.

Using the same approach, we also single cell–sorted and expanded DQ6-, DR0401-, and DR0301-restricted GAD65/IGRP tetramer+CD45RACD127+CD25 Teffs of multiple specificities. We then analyzed the cytokine responses of DR1501-restricted GAD65/IGRP-specific Treg clones and DR1501-, DQ6-, DR0301-, and DR0401-restricted GAD65/IGRP-specific Teff clones elicited by peptide-specific stimulation in vitro. In accord with previous studies (47), DR1501-restricted GAD65/IGRP antigen–specific Treg clones were incapable of producing cytokines such as interleukin-2 (IL-2), IL-4, interferon-γ (IFN-γ), IL-17, and IL-21 (fig. S5). A higher percentage of DR1501-restricted GAD65/IGRP antigen–specific Teffs produced IL-10 compared with DR-restricted Teffs derived from DR0401-DQ8 or DR0301-DQ2 individuals (Fig. 5, A and B), suggesting that these IL-10–producing DR1501-restricted Teffs could contribute to the HLA-linked protection in individuals with the DR1501-DQ6 haplotype. In addition, more DQ6-restricted Teffs produced IL-10 compared with DR3-restricted Teffs. Although the observed percentage of DQ6-restricted Teffs that could produce IL-10 was also higher than DR4-restricted Teffs (3.5 versus 2.5%), that difference did not reach statistical significance. In addition, Teffs from both the protective haplotype and susceptible haplotypes produced IFN-γ and IL-4.

Fig. 5 Cytokine profiles of GAD65- and IGRP-specific CD4+ Teff clones.

(A) Strategy for analyzing T cell function of tetramer+ cells in vitro. Cells were stimulated with an irrelevant negative control peptide (Flu MP97–116) or a specific GAD65 or IGRP peptide and then intracellularly stained for IL-2, IL-4, IL-10, IFN-γ, IL-21, and IL-17A. (B) Comparison of cytokine levels between T1D high-risk (25 DR0301 clones and 33 DR0401 clones) and T1D-protective (30 DR1501 clones and 30 DQ6 clones) groups (specificities of all clones are in table S1). Shown is the mean ± SEM. Welch’s t test was used. N.S., P ≥ 0.05; *P < 0.05; and **P < 0.01.

Islet antigen–specific Tregs suppressed autoreactive Teff responses

In the past, the suppressive capacity of Tregs has commonly been assayed upon nonspecific stimulation (e.g., anti-CD3/CD28 beads) (48, 49). Because our goal was to evaluate Treg suppression in an antigen-specific fashion, we developed an assay in which the proliferation of islet antigen–specific Teffs was examined in the presence of HLA-matched antigen-presenting cells and islet antigen–specific Tregs upon specific peptide stimulation. Using this approach, we observed that (i) DR1501-restricted GAD65137–156-specific Tregs could inhibit the proliferation of DR1501-restricted GAD65137–156-specific Teffs in the presence of the GAD65137–156 peptide and (ii) DR1501-restricted IGRP225–244 Tregs could inhibit the proliferation of IGRP225–244-specific Teffs in the presence of IGRP225–244 peptide (Fig. 6, A and B). Significant suppression was observed with a Treg/Teff ratio of 1:8, demonstrating the potency of antigen-specific Tregs.

Fig. 6 Antigen-specific suppression of DR1501-restricted autoreactive Teffs in vitro by DR1501-restricted islet antigen–specific Tregs.

A total of 10,000 T responders from GAD65- or IGRP-specific Teff clones, cultured with cells from a Treg clone recognizing the same peptide epitope, at several Treg/Teff ratios. DR1501 PBMCs were used as antigen-presenting cells. GAD65 or IGRP peptide was added, and the proliferation of responder Teffs was analyzed after 6 days. (A) Representative fluorescence-activated cell sorting (FACS) analysis. (B) Percent suppression by antigen-specific Treg clones. Each circle represents a distinct Treg clone (filled circles, GAD65137–156 clones; blank circles, IGRP225–244 clones). Shown is the mean ± SD. Paired t test was used. *P < 0.05; **P < 0.01; and ****P < 0.0001.

DQ8-restricted islet antigen–specific CD4+ T cells play an important role in the T1D pathogenesis (5053) and could be less subject to regulation. We hypothesized that to achieve a dominant protective effect, DR1501-restricted Tregs would need to regulate DQ8-restricted Teffs in DR1501-DQ6/DR0401-DQ8 heterozygous individuals. To test this hypothesis, we examined the suppressive capacity of DR1501-restricted GAD65137–156-specific Tregs on DQ8-restricted GAD65250–266-specific Teffs in the presence of DR1501-DQ6/DR0401-DQ8 antigen-presenting cells and GAD65137–156 and GAD65250–266 peptides. As shown in Fig. 7 (A and B), DR1501-restricted GAD65137–156 Tregs inhibited the proliferation of DQ8-restricted GAD65250–266 Teffs in the presence of both GAD65137–156 and GAD65250–266 peptides. However, suppression was not observed in the absence of Tregs or the GAD65137–156 peptide.

Fig. 7 Suppression of DQ8-restricted GAD65 Teffs in vitro by DR1501-restricted GAD65 Tregs.

A total of 10,000 T responders from DQ8-restricted Teff clone, cultured with cells from a DR1501-restricted Treg clone, at several Treg/Teff ratios. DR1501-DQ8 heterozygous PBMCs were used as antigen-presenting cells. DQ8 Teff recognizes peptide 1 (GAD65250–266), and DR1501 Treg recognizes peptide 2 (GAD65137–156). (A) Representative FACS analysis. The number within each panel indicates the percentage of proliferating cells. (B) Cumulative percent suppression by different antigen-specific Treg clones (n = 4). (C) Illustration of the transwell coculture system for suppression assay. (D) Cumulative percent suppression of responder cells at upper well and lower well (n = 4). (E) Cumulative percent suppression by Treg clones with addition of isotype control or anti-human IL-10 monoclonal antibody or exogenous human IL-2 (n = 4). Shown is the mean ± SD. Paired t test was used. N.S., P ≥ 0.05; *P < 0.05; **P < 0.01; and ***P < 0.001.

To investigate the functional mechanism by which islet-specific CD25+CD127 Tregs exerted their suppressive function, we used transwell experiments to examine whether the suppressive function of these Tregs was mediated by cell-to-cell contact or by soluble mediators. Although cell-to-cell contact provided maximum suppressive function, suppression was also observed in the absence of cell-to-cell contact (Fig. 7, C and D, and fig. S7). The cell contact–independent suppressive mechanism was further examined by blocking IL-10. In agreement with our experimental data indicating that the DR1501 Tregs did not produce measurable levels of IL-10 (fig. S5), addition of anti–IL-10 antibody did not abrogate the suppressive function (Fig. 7E and fig. S8). In contrast, addition of excess IL-2 partially restored the proliferation of Teffs (Fig. 7E and fig. S8). These results suggested either that IL-2 consumption by Tregs deprived Teffs of IL-2 or that addition of IL-2 helps the Teffs to overcome the suppressive effect of other unidentified cell contact–dependent or cell contact–independent pathway. In summary, these series of experiments revealed that the DR1501-restricted GAD65-specific Tregs used both cell contact–dependent and cell contact–independent mechanisms to suppress Teffs.

Increased frequencies of islet antigen–specific Teffs and Tregs in individuals with protective haplotype by CD154/CD137 up-regulation assay

The results we obtained through these tetramer-based assays were limited to T cells that recognized two T1D-associated autoantigens GAD65 and IGRP. To confirm that comparatively high frequencies of islet antigen–specific Teffs and Tregs are present in individuals with DR1501-DQ6 haplotype in a more comprehensive fashion, we applied recently developed CD154/CD137 up-regulation assays upon antigen-specific stimulation (47, 54) to identify and enumerate islet antigen–specific Tregs and Teffs. This approach used overlapping peptides derived from GAD65, IGRP, preproinsulin (PPI), and zinc transporter 8 (ZnT8) as the stimulating antigens. Peptides from PPI and ZnT8 were included because these proteins are considered to be major autoantigens (5557). For these experiments, Teffs were defined as CD4+CD154+CD69+ T cells, and Tregs were defined as CD4+CD154CD137+GARP+CD127CD25+ (Fig. 8A). Consistent with the results of our tetramer-based experiments, higher frequencies of islet antigen–specific CD4+ Teffs and Tregs were observed in DR1501-DQ6 individuals compared with DR0401-DQ8 individuals (Fig. 8B).

Fig. 8 Antigen-reactive T cell enrichment assay with islet antigens including GAD65, IGRP, PPI, and ZnT8.

(A) Strategy for identifying islet antigen–specific Teffs and Tregs ex vivo. (B) The frequencies of islet antigen (Ag)–specific CD4+ (left) and CD45RACD4+ memory (right) Teffs and Tregs were compared between T1D high-risk (DR0401-DQ8, n = 10) and T1D-protective (DR1501-DQ6, n = 16) groups. Outlier data points, as identified by ROUT (Q = 1%, GraphPad Prism 7), were removed for the final analysis. Shown is the mean ± SEM. Welch’s t test was used. *P < 0.05 and **P < 0.01.


The mechanism by which the DR1501-DQ6 haplotype confers dominant protection in T1D is unclear. Previous studies have focused on the mechanism of epitope stealing mediated by DQ6. Those studies demonstrated that for individuals with the DR1501-DQ6/DR0401-DQ8 genotype, DQ6 HLA molecules on the surfaces of antigen-presenting cells could compete with DQ8 molecules for identical antigenic peptides and alter DR0401-DQ8–restricted immune responses. In support of this hypothesis, Eerligh et al. (29) demonstrated that the responses of DQ8-restricted insulin-specific T cell clones switched from proinflammatory to anti-inflammatory phenotype when antigen-presenting cells used to generate the clone were switched from a DQ8/8 to a DQ6/8 genotype. Van Lummel et al. (28) showed that binding of insulin B 6–23 to DQ8 was significantly decreased in the presence of DQ6, concluding that the peptide was prevented from binding to DQ8 because it bound preferentially to DQ6. Although our experiments did not directly address whether epitope stealing occurred, this current study did identify multiple DR1501- and DQ6-restricted GAD65 and IGRP epitopes (Table 1), three of which overlapped with epitopes that could be presented by DR0401, DR0301, or DQ8. Specifically, the DR1501- and DQ6-restricted GAD65553–572 epitope could be presented by DR0401 (33), and the DR1501-restricted IGRP225–244 epitope could be presented by DR0301 (34). The DQ6-restricted GAD65241–260 epitope also partially overlapped with a previously reported DQ8-restricted GAD65250–266 epitope (51, 58). Thus, the presence of the DR1501 and DQ6 alleles could potentially influence antigen presentation by DR0301, DR0401, and DQ8. However, additional studies will be needed to formally address this question. In addition, cytokines from DR1501- and DQ6-restricted T cell responses should also deviate the immune responses elicited from the susceptible DR- and DQ-restricted T cells. However, the extent of epitope overlap between the protective and susceptible alleles is limited. We reasoned that the DR1501-DQ6 haplotype may have additional protective effects, thereby limiting the risk of developing T1D in individuals who have a susceptible haplotype.

In the current study, we applied both class II tetramer staining and CD154/CD137 up-regulation assays to examine the frequencies of islet antigen–specific CD4+ T cells and Tregs in healthy individuals with the protective haplotype, susceptible haplotypes, or a neutral haplotype. We chose to focus on healthy individuals rather than individuals with T1D because we seek to understand the attributes conferred by the HLA, independently of other factors, such as hyperglycemia, that arise as a consequence of T1D onset. On the basis of class II tetramer staining, the frequencies of DR1501-restricted GAD65 autoreactive Teffs and Tregs in individuals with DR1501-DQ6 haplotype were significantly higher than those of DR0401-, DR0301-, and DR0701-restricted GAD65-specific Teffs and Tregs in individuals with DR0401-DQ8, DR0301-DQ2, and DR0701-DQ2.2 haplotypes. Frequencies of DR1501-restricted GAD65-specific CD4+ T cells were also found to be higher in comparison with DQ6-restricted GAD65-reactive cells. The frequencies of DR1501-restricted IGRP-autoreactive CD4+ and Tregs were also significantly higher in most comparison with IGRP-specific cells in DR0401-DQ8, DR0301-DQ2, and DR0701-DQ2.2 individuals. The same relative abundance of islet antigen–specific Teffs and Tregs in individuals with the protective DR1501-DQ6 haplotype in comparison with individuals with the DR0401-DQ8 haplotype was more comprehensively confirmed using a CD154/CD137 up-regulation assay.

The importance of Tregs in T1D is well recognized. For example, it was shown that Tregs in T1D individuals are defective in their suppressive capacity (37, 38). It has also been reported that Teffs in T1D are resistant to suppression (59). Although most studies showed no differences in the frequencies of Tregs between healthy and T1D individuals, none of these addressed the frequencies of islet antigen–specific Tregs (37, 38, 60, 61). The difference in frequencies of GAD65 and IGRP Tregs among the protective DR1501-DQ6 haplotype group and the susceptible and neutral HLA haplotype groups appears to be islet antigen specific, because there was no difference in the overall percentage of CD25+CD127 Tregs in the total CD4+ T cell populations and the frequencies of Flu B hemagglutinin (HA)–specific Tregs between these different groups were similar (fig. S6, A and B). This supports a possible role of islet antigen–specific Treg in mediating the protective effect in individuals with the DR1501-DQ6 haplotype and implies that the protective effect is both antigen and HLA specific. Through our methodology, we also highlighted the successful cloning of naturally derived epitope-specific Tregs from human individuals. In this study, the Treg epitopes studied were identical to Teff epitopes. Because the isolation of these Tregs was biased by the use of tetramer reagents, it remains unclear whether epitopes recognized by Tregs and Teffs are essentially identical. It also remains unclear whether the Tregs detected in our assays are pTregs or tTregs. However, the Treg clones isolated were Helios+, raising the possibility that they could be thymically derived (62). It will also be of interest to compare the suppression potency of antigen-specific Tregs isolated from healthy individuals with protective and susceptible alleles as well as from T1D individuals.

The presence of higher frequencies of epitope-specific Tregs in an individual with a protective haplotype was also reported for Goodpasture disease (63). In that study, individuals with a protective haplotype had higher frequencies of Tregs and lower frequencies of Teffs compared with individuals with susceptible haplotype, leading to significant differences in the Treg/Teff ratio between the two groups. In our current study, we also observed a higher frequency of Teffs in individuals with a protective haplotype compared with those with susceptible haplotypes, leading to similar Treg/Teff ratios between the protective and susceptible groups.

The high frequency of autoreactive Teffs in DR1501-DQ6 individuals is in contrast to the conventional reasoning that the frequency of CD4+ autoreactive T cells is higher in individuals with the susceptible haplotypes. However, we also observed that a higher percentage of T cells derived from DR1501-restricted autoreactive Teffs produced IL-10 compared with the DR0401- and DR0301-restricted cells. The propensity of the DR1501-restricted cells to produce IL-10 could be taken to suggest deviation toward a Tr1-like lineage or a reduced potential for sustaining autoreactivity. In total, our data suggest that protection mediated through the DR1501-DQ6 haplotype is not due to the absence of autoreactive Teffs. Rather, the high abundance of DR1501-restricted islet-specific Tregs and the altered cytokine profiles of islet-specific Teffs imply a multifaceted role of these cells in conferring dominant protection.

Hauben et al. (64) have put forth a hypothesis of “beneficial autoimmunity,” in which robust but adequately regulated autoreactive T cell responses can facilitate the clearance of damaged tissues and prevent the subsequent self-perpetuating inflammation and autoimmunity. In this model, cytokines and chemokines from autoreactive T cells will activate the innate immune cells for clearance of dead cells, which would be a source of autoantigens that propagate the inflammatory and autoimmune responses. In contrast, a weak initial CD4+ T cell response, which is unable to activate the innate immune cells to clear the damaged tissues, will potentially lead to chronic inflammation and autoimmunity.

The observation of high frequencies of islet-specific Teffs and Tregs in the protective group supports this beneficial autoimmunity model. In this hypothetical scenario, β cell injury will lead to trafficking of DR1501-restricted autoreactive T cells into the injured tissue site. Cytokines/chemokines from the DR1501-restricted Teffs then activate the innate cells that promote the clearance of β cell debris as a first step to restore immune homeostasis. At the same time, activation of DR1501-restricted islet antigen–specific Tregs and production of IL-10 from DR1501-restricted Teffs will act together to down-regulate the DR1501-restricted Teff responses and the innate responses from becoming overexuberant and harmful to the host. In addition, both FOXP3+ Tregs and IL-10 from Teffs are capable of regulating other islet antigen–specific T cells restricted by other class I and class II alleles, including DQ8-restricted autoreactive T cells, in their immediate environment. Hence, a high frequency of Teffs that can be adequately restrained by Tregs could be beneficial to the host in tissue healing.

Our study demonstrates that the DR1501-DQ6 haplotype appears to confer protection through multiple pathways. First, DR1501-restricted islet-specific FOXP3+ Tregs and IL-10–producing Tr1-like cells can down-modulate the pathogenic immune responses, beneficially altering the islet milieu. There is also the possibility that IFN-γ production from DR1501-restricted Teffs can provide beneficial autoimmunity, facilitating the elimination of autoantigens through clearance of damaged islets in an indirect fashion. In total, these data show that individuals with the DR1501-DQ6 haplotype have unique T cell repertories that can act through different mechanisms to achieve dominant protection. Additional studies in examining autoreactive T cells restricted by additional susceptible, neutral, and protective DR and DQ alleles should be fruitful in dissecting the cellular pathways that promote disease protection.


Study design

The goal of this study was to determine the role of DR1501-DQ6–restricted islet antigen–specific CD4+ T cells in conferring dominant protection in T1D. DR1501- and DQ6-restricted GAD65- and IGRP-specific CD4+ T cell epitopes were identified by TGEM. Frequencies of islet antigen–specific Teffs and Tregs in healthy individuals with DR1501-DQ6, DR0401-DQ8, DR0301-DQ2, and DR0701-DQ2.2 haplotypes were analyzed by ex vivo staining with class II tetramers. GAD65- and IGRP-specific CD4+ T cells from individuals with different haplotypes were cloned, and their cytokine profiles were examined. The suppressive function of DR1501-restricted Treg clones was examined by an in vitro suppression assay in an antigen-specific fashion. Besides GAD65 and IGRP, other islet antigens including PPI and ZnT8 peptide pools were included in CD154/CD137 assays to comprehensively identify the islet antigen–specific CD4+ Teffs and Tregs in individuals with the DR1501-DQ6 or DR0401-DQ8 haplotype.


Healthy individuals with specific HLA class II haplotypes of interest were recruited at Benaroya Research Institute at Virginia Mason under a study approved by the Benaroya Research Institute Institutional Review Board. All individuals were recruited with written informed consent. A total of 16 DR1501-DQ6 individuals were included in this study. Participants who did not have DR1501-DQ6 but had other HLA haplotypes of interest, including 10 with DR0401-DQ8, 6 with DR0301-DQ2, and 7 with DR0701-DQ2.2, were also recruited to this study.

Peptides and tetramers

The islet antigens in our study included GAD65, IGRP, PPI, and ZnT8 proteins. Overlapping peptides that were 20 amino acids in length with 12–amino acid overlap that covered the entire protein for each of the four antigens mentioned were ordered from Mimotopes. Peptides were loaded onto the specific HLA class II protein to generate tetramers as previously described (35, 65).

Tetramer-guided epitope mapping

TGEM was performed as described (34, 35) to identify MHC class II–restricted peptide epitopes for the islet antigens GAD65 and IGRP. Briefly, GAD65- and IGRP-derived peptides were divided into pools of five peptides each. Each pool was added to an individual well in a 24-well plate, with 4 × 106 to 5 × 106 PBMCs isolated from a donor of interest. Cells were cultured for 14 to 19 days and then stained with pooled peptide-loaded tetramers. Cells from wells with a positive staining result were screened again by staining with each tetramer loaded with individual peptides that belong to that particular peptide pool. The specific peptide for the loaded tetramer that gave a positive staining was considered to be an epitope. The specificity of tetramer staining was confirmed by single-cell cloning of the tetramer-positive cells and then restaining the T cell clone with the tetramer used for sorting.

Ex vivo peptide–MHC II tetramer analysis

PBMCs were isolated from whole-blood samples by Ficoll underlay. Ex vivo tetramer staining was then performed as described previously (51, 66). Briefly, 90 million PBMCs were treated with dasatinib (50 nM) at 37°C for 5 min, washed once, and then divided into three tubes. Cells were then were stained with phycoerythrin (PE)–, PE-Cy5–, or PE-Dazzle 594–labeled tetramers (20 μg/ml for each tetramer) at room temperature for 1.5 to 2 hours. The tetramers used were selected to match the relevant HLA of interest as listed in Table 1. Cells were then washed and incubated with anti-PE MicroBeads (Miltenyi Biotec) at 4°C for 10 min. After reserving a precolumn fraction (used to estimate the total number of CD4+ T cells within the sample), tetramer-positive cells were enriched using MS column (Miltenyi Biotec) and then stained with other surface antibodies including anti-CXCR5–allophycocyanin (APC)–Cy7 (clone J252D4), anti-CCR4–PerCP-Cy5.5 (clone L291H4), anti-CXCR3–BUV395 (clone 1C6/CXCR3, BD Biosciences), anti-CCR6–BV650 (clone G034E3), anti-CD45RA–AF700 (clone HI100), anti-CD4–BUV737(clone SK3, BD Biosciences), anti-CD127–BV711 (clone A019D5), and anti-CD25–BV785 (clone M-A251), and a dump channel consisting of anti-CD14–BV510 (clone M5E2), anti-CD19–BV510 (clone SJ25C1), anti-CD56–BV510 (clone HCD56), and Fixable Viability Stain 510 (BD Biosciences). These surface-stained cells were treated with Nuclear Transcription Factor Buffer Set and then intracellularly stained with anti-FOXP3–FITC (fluorescein isothiocyanate) (clone 206D) and anti-Helios–APC (clone 22F6) as necessary. All reagents mentioned above were purchased from BioLegend except otherwise noted. Cells in all three tubes were then pooled together and analyzed with a BD LSRII; data collected were analyzed by FlowJo 10.4.2. Frequencies were calculated by dividing the number of tetramer-positive cells in the bound fraction by the number of total CD4+ T cells, as determined by analyzing the precolumn sample.

Single-cell cloning of peptide–MHC II tetramer-positive cells

Antigen-specific CD4+ T cells were magnetically enriched as described above and then single cell–sorted by BD FACSAria Fusion. Individual dump gate negative CD4+CD45RAtetramer+ cells were sorted at single-cell purity directly into individual wells of a 96-well plate and then expanded with 0.1 million irradiated human PBMCs as feeder cells and phytohemagglutinin (PHA) (1 μg/ml; Remel) in 100 μl of T cell medium (TCM; consisting of RPMI with 10% pooled human serum and 1% sodium pyruvate, glutamine, and penicillin/streptomycin; Invitrogen) per well as described (67). Recombinant human IL-2 (PeproTech) was added to the culture the next day to reach the final concentration of 300 IU/ml (41) for CD127CD25+CD4+ Tregs or 10 IU/ml for CD127+CD25CD4+ Teffs. The cells were incubated at 37°C, 5% CO2 for 2 to 3 weeks and fed with fresh TCM and IL-2 as necessary. After in vitro expansion, the antigen specificity of T cell clones was confirmed again by tetramer staining.

Cytokine assay

After stimulation and expansion, T cell clones were rested in TCM without PHA or IL-2 for 3 days. For each well, about 0.1 million cells in 100 μl of TCM were cultured with phorbol 12-myristate 13-acetate and ionomycin (T Cell Activation Cocktail, BioLegend), antigen-specific peptide (10 μg/ml; Mimotopes), or irrelevant peptide (10 μg/ml; influenza matrix protein MP97–116, Mimotopes; see Table 1) as a negative control. Cells were stimulated for 6 hours and treated with brefeldin A (5.0 μg/ml) for the last 3 hours. Cell surface staining was performed first, followed by fixation, permeabilization, and then staining of intracellular cytokines including IL-2–BV650 (clone MQ1-17H12), IL-4–BV605 (clone MP4-25D2), IL-10–PE-Cy7 (clone JES3-9D7), IFN-γ–BV421 (clone 4S.B3), IL-21–APC (clone 3A3-N2), and IL-17A–PE (clone BL168). All reagents mentioned above were purchased from BioLegend.

Treg in vitro suppression assay

The regulatory function of the antigen-specific Treg clones was examined in an antigen-specific fashion by using a suppression assay that was adapted from a previously described assay (48, 49). Briefly, 10,000 cells of the islet antigen–specific Teff clone of interest were used as responder cells and labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE) or using the CellTrace Violet Cell Proliferation Kit (Invitrogen). The responder cells were cultured with Tregs at Treg/Teff ratios of 0:1, 1:8, and 1:1. HLA-matched PBMCs (100,000) were pulsed with specific antigenic peptide at 10 μg/ml, irradiated at 50 Gy, and used as antigen-presenting cells. In some experiments, anti–IL-10 antibody (10 μg/ml, clone JES3-19F1, BD Biosciences), its isotype control (clone R35-95, BD Biosciences), or human IL-2 (20 IU/ml, Roche) was added. After antigen-specific stimulation for 6 days, the proliferation of responder cells was analyzed with an LSRII flow cytometer (BD Biosciences). The percentage of suppression was calculated by using the following formula: % Suppression = (proliferation of responder cell cultured alone − proliferation of responder cell cocultured with Tregs)/proliferation of responder cell cultured alone. For transwell experiments, Millicell-96 cell culture insert plates (Millipore) were used. Ten thousand CFSE-labeled DQ8-restricted GAD65-specific Teffs and 100,000 irradiated PBMCs from a DR1501/DR4-DQ8 individual were placed in both the upper and lower chambers. DR1501-restricted GAD65-specific Tregs were added to the upper chamber only at different Treg/Teff ratios. Both Teffs and Tregs were stimulated with relevant peptides (10 μg/ml). After 6 days of culture, proliferation of Teffs in both upper and lower chambers was analyzed in parallel to show the effect of cell-cell contact on Treg suppression.

CD154/CD137 assays for detection of islet-specific Teffs and Tregs

Assays for the detection of islet-specific CD4+ Teffs and Tregs were performed essentially as described (47, 54, 67). Briefly, 30 million PBMCs in 2 ml of TCM were stimulated for 16 hours with T1D peptide pools (2 μg/ml for each peptide) derived from GAD65, IGRP, PPI, and ZnT8 proteins in the presence of anti-CD40 (1 μg/ml, clone HB-14, Miltenyi Biotec). Cells were then stained with PE-conjugated anti-CD154 (clone 5C8, Miltenyi Biotec) and PE-Cy7–conjugated anti-CD137 (clone 4B4-1). A 1/10th fraction of the cells was saved, and the rest of the CD154-PE+/CD137-PE-Cy7+ cells were magnetically enriched by anti-PE MicroBeads and MS column (Miltenyi Biotec). The enriched cells were stained with additional surface antibodies including anti-CD69–PE-Cy5 (clone FN50), anti-GARP–APC (clone 7B11), anti-CD45RA–AF700 (clone HI100), anti-CD4–BUV737 (clone SK3, BD Biosciences), anti-CD127–BV711 (clone A019D5), and anti-CD25–BV785 (clone M-A251), and a dump channel consisting of anti-CD14–BV510 (clone M5E2), anti-CD19–BV510 (clone SJ25C1), anti-CD56–BV510 (clone HCD56), and Fixable Viability Stain 510 (BD Biosciences). All antibodies mentioned above were purchased from BioLegend except otherwise noted. The frequency of Teffs was calculated by using the formula F = n/N, where n designates the number of CD154+CD69+ cells in the bound fraction after enrichment and N is the total number of CD4+ T cells, which was calculated as 10 × the number of unenriched cell. The frequency of Tregs was calculated by using the same formula, where n is designated as the number of CD154CD137+GARP+CD127CD25+CD4+ T cells.

Statistical analysis

GraphPad Prism software (version 7.05) was used for data analysis. Value is the mean ± SEM (for population) or SD (for Treg in vitro suppression assay). Welch’s t test or paired t test was used. N.S., P ≥ 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; and ****P < 0.0001. Outliers were identified by ROUT (robust regression and outlier removal; Q = 1%) and then removed to prevent outlier bias.


Fig. S1. Identification of DR1501-restricted GAD65 epitopes by TGEM.

Fig. S2. Tetramer staining of GAD65 and IGRP antigen–specific cells.

Fig. S3. GAD65137–156 is the dominant DR1501-restricted GAD65-specific T cell epitope.

Fig. S4. GAD65137–156 is a naturally processed epitope.

Fig. S5. Cytokine profiles of DR1501-restricted Treg and Teff clones.

Fig. S6. Total and Flu-specific CD4+CD127CD25+ Tregs among different haplotype groups.

Fig. S7. Suppression of DQ8-restricted GAD65 Teffs with DR1501-restricted Tregs in transwell chamber assay.

Fig. S8. Effects of anti–IL-10 or exogenous IL-2 in suppression assays.

Table S1. Specificities of DR0301-, DR0401-, DR1501-, and DQ6-restricted Teff clones.

Table S2. Specificities of DR1501-restricted Treg clones.

Data file S1. Raw data file (Excel spreadsheet).


Acknowledgments: We thank C. Cousens-Jacobs for administrative support and preparation of this manuscript. We also thank the Benaroya Research Institute Tetramer Core Laboratory for producing all of the class II monomers used in this study. J.Y. is currently employed by Cs-Bay Therapeutics Inc. (8000 Jarvis Ave. #208, Newark, CA 94560, USA). Funding: This work was supported by NIH grants DP3 DK097653 and DK106909 and Leona M. and Harry B. Helmsley Charitable Trust grant 2018PG-T1D036. Author contributions: X.W., J.Y., H.R., and W.W.K. designed the study. X.W., J.Y., and E.J. performed the experiments. I.-T.C. provided critical reagents. W.W.K. supervised the work. X.W., E.J., and W.W.K. wrote and edited the manuscript. All authors reviewed and edited the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials. The tetramer reagents used in this work can be obtained from the Benaroya Research Institute Tetramer Core Laboratory ( through a material transfer agreement. The availability of individual T cell clones used in this study may be limited.

Stay Connected to Science Immunology

Navigate This Article