Proinsulin-Specific, HLA-DQ8, and HLA-DQ8-Transdimer–Restricted CD4⁺ T Cells Infiltrate Islets in Type 1 Diabetes

Type 1 diabetes (T1D) is an autoimmune disease caused by the CD4⁺ and CD8⁺ T-cell-mediated destruction of pancreatic insulin-producing β-cells (1). β-Cell destruction leads to primary insulin deficiency, which is treated by exogenous insulin therapy, and currently there is no cure. The pathogenesis of T1D has been well characterized using the NOD mouse model, but the immune basis of T1D in humans is less clear.

Genetic association studies have provided powerful insights into the etiology of human T1D (2,3). The HLA class II region has the strongest impact on risk of T1D. Some HLA alleles—DQB1*06:02 for example—dominantly protect against T1D (4). In contrast, of all alleles, HLA-DQ2 (DQA*05:01, DQB*02:01) and DQ8 (DQA*03:01, DQB*03:02) confer the greatest risk of developing T1D (5). Remarkably, individuals heterozygous for HLA-DQ2 and HLA-DQ8 are at greater risk of developing T1D than those with either HLA-DQ2 or -DQ8 alone (6). The basis of this observation is unclear. However, HLA-DQ2 or -DQ8 heterozygous antigen-presenting cells (APCs) can form two extra functional HLA molecules: a DQ8 transdimer composed of the DQ2α chain paired with the HLA-DQ8β chain (DQA*05:01; DQB*03:02) and, similarly, a DQ2 transdimer in which the DQ8β chain pairs with DQ2α (DQA1*03:01; DQB1*02:01) (7). These transdimers may promote β-cell autoimmunity by presenting unique diabetogenic epitopes, or the high density of T1D-promoting HLA molecules (DQ2, DQ8, DQ2trans, and DQ8trans) may promote autoimmune CD4⁺ T-cell responses against β-cell antigens (7,8). Because the epitopes presented by HLA-DQ2/-DQ8 transdimers have not been identified, these possibilities remain speculative. Nonetheless, the function of HLA-DQ is to present antigens to CD4⁺ T cells; hence these genetic associations are strong evidence that CD4⁺ T-cell responses against β-cell antigens play a pivotal role in the development of T1D in humans.

Because insulin-specific autoantibodies were detected before the onset of T1D (9), insulin and its precursor proinsulin have been candidate autoantigens in T1D. It is now clear that T-cell responses to (pro)insulin are essential for the development of T1D in the NOD mouse (10,11), but the role of (pro)insulin in human T1D has not been confirmed. Genetic association studies have implicated proinsulin because a T1D susceptibility locus, IDDM2, maps to a variable number of tandem repeats (VNTR) upstream of the insulin gene (12,13). This polymorphism is believed to modulate proinsulin expression in the thymus, affecting central tolerance (12–14). This observation provides circumstantial evidence that proinsulin plays a crucial role as an autoantigen in T1D. Many studies have attempted to detect proinsulin-specific CD4⁺ T-cell responses in the peripheral blood mononuclear cells (PBMCs) of patients with T1D and healthy control subjects (15). Using sensitive methods capable of detecting rare T cells, some investigators detected weak responses to proinsulin peptides (16). However, these CD4⁺ T cells could not be analyzed in detail. Furthermore, T cells isolated from the pancreatic lymph nodes of deceased organ donors who had T1D were reported to be insulin specific but HLA-DR4 restricted (17).

These genetic associations between HLA and insulin alleles predict that HLA-DQ2- and HLA-DQ8-restricted, proinsulin-specific CD4⁺ T cells will infiltrate human islets, where they play a pathogenic role in T1D. Although predicted, these CD4⁺ T cells have never been isolated from human islets, leaving a crucial piece missing from the chain of autoimmune causation in T1D. This gap in our knowledge has led some to question about the autoimmune basis of human T1D (18,19).

Here we show that CD4⁺ T cells isolated from within the islets of an organ donor who had T1D recognize proinsulin epitopes presented by HLA-DQ8 and HLA-DQ8 transdimers. This work reveals how molecules—HLA-DQ8 and proinsulin—implicated by genetic association studies lead to autoimmune response against human β-cells and provides new evidence to support the notion that human T1D is an autoimmune disease.

The isolation of pancreatic islets from deceased organ donors was approved by the St. Vincent’s Hospital Human Research Ethics Committee (approval no. SVH HREC-A 011/04). Collection of peripheral blood samples from patients with T1D was approved by Southern Health Human Research Ethics Committee-B (approval no. 12185B). Collection of blood from healthy donors was approved by St. Vincent’s Hospital Human Research Ethics Committee (approval no. SVH HREC-A 135/08).

After obtaining informed consent from the next of kin, the pancreas was removed from a brain-dead donor and sent to the Tom Mandel Islet Transplant Program. Islets were purified by intraductal perfusion and digestion of the pancreas with collagenase followed by purification using Ficoll density gradients, as previously described (20). Islet purity was assessed by light microscopy after staining with dithizone (Sigma Chemicals, St. Louis, MO). Thirty thousand islet equivalents were isolated from the tail of the pancreas. Approximately 1,000 islets were cultured in CMRL 1066 overnight then stained to detect T cells by flow cytometry. All the remaining islets were cultured directly after isolation in a 24-well plate in 1.0 mL of RPMI medium/5% pooled human serum containing interleukin (IL)-2 (20 U/mL) and IL-15 (5 ng/mL) for 7–10 days. After this time a halo of lymphocytes was visible surrounding 3–4 islets. Cells from wells containing growing T cells were collected and stained with anti-CD3-PerCP (UHCTI), CD4-phycoerythrin (OKT4), and anti-CD8α-Alexa Fluor 647 (OKT8). Dead cells were excluded with propidium iodide staining. CD3⁺/CD4⁺ cells were cloned by sorting a single CD4⁺ T cell into wells of a 96-well plate containing irradiated allogeneic PBMCs, anti-CD3, and cytokines, as described elsewhere (21,22). After 2–3 weeks, growing clones were collected and subjected to another round of CD3-stimulated growth in the presence of IL-2 (20 U/mL) and IL-4 (5 ng/mL), as described previously (21,22). In most cases this expansion yielded >100 million cloned CD4⁺ T cells. Aliquots of five million expanded T cells were cryopreserved in 10% DMSO/FCS, as described previously (23), and stored over liquid nitrogen until required.

Specimens from the head of the pancreas were taken before perfusion with collagenase and were snap frozen in optimal cutting temperature compound (Sakura Finetek, Torrance, CA) Frozen sections were fixed in acetone for 5 min at room temperature and then washed in PBS. Endogenous peroxidases were quenched by incubating sections with 0.3% hydrogen peroxide (Merck, Whitehouse Station, NJ) for 15 min. Nonspecific protein–binding sites were blocked by incubating sections with 2.5% milk powder and 10% FCS in PBS for 15 min. Sections were incubated with mouse anti-human CD4 or guinea pig anti-insulin and then horseradish peroxidase (HRP)-rabbit antimouse IgG or HRP–rabbit anti–guinea pig antibodies (Dako, Glostrup, Denmark), followed by staining with a Vectastain ABC Kit (Vector Laboratories, Burlingame, CA) following the manufacturer’s instructions. HRP staining was detected with diaminobenzidine tablets (Sigma-Aldrich, St. Louis, MO), and sections were counterstained with hematoxylin. To confirm labeling specificity, tissue sections were incubated with isotype antibodies, no primary antibody, or diaminobenzidine alone (controls). Sections were viewed on a Leitz Laborlux K microscope and photos were taken with an Olympus CKX41 camera.

Peptides were synthesized by GL Biochem (Shanghai, China) using 9-fluorenylmethoxycarbonyl chemistry. Peptides were purified by reversed-phase high-performance liquid chromatography to at least 85% purity and lyophilized. Peptides were reconstituted in 40% acetonitrile, 0.5% acetic acid, and water to 5 mmol/L, aliquoted, and stored at −80°C. A complete list of peptides is shown in Supplementary Tables 4–9.

Cloned CD4⁺ T cells were thawed and used directly in functional assays. To identify the epitopes recognized by the cloned CD4⁺ T cells, they were incubated with a class II HLA-matched Epstein Barr virus–transformed B-cell line (KJ; HLA-DRB1*03:01, 04:04; DQB1*02:01, 03:02) and synthetic peptides that mimic the entire sequence of human proinsulin or a total of 26 peptides reported to be epitopes recognized by CD4⁺ T cells from patients with T1D (Supplementary Tables 4–9). Cloned CD4⁺ T cells (10–50,000/well) were cultured with 5–10,000 APCs with and without peptides, as indicated in the figures. The response of a T cell to antigen was measured as interferon (IFN)-γ secretion into the culture media. The IFN-γ concentration in the culture media was determined by ELISA (BioLegend, San Francisco, CA). In preliminary screening experiments an IFN-γ standard curve was omitted; in these cases the results are plotted as optical densities at 450 nm.

HLA restriction was determined in two steps, as described previously (24–26). First, monoclonal antibodies specific for HLA-DR (clone L243), -DP (clone B7/21), and -DQ (clone SPV-L3) were added to the peptide-stimulated T-cell cultures to a final concentration of 1–10 μg/mL. Second, the HLA alleles were determined using a panel of T2 cells transduced with a retrovirus encoding for HLA-DQA and -DQB alleles (specific alleles are indicated in the figure legends).

RNA was extracted from one to two million cloned T cells using an RNA extraction kit (Macherey Nagel, Duren, Germany) and converted to cDNA using SMARTScribe Reverse Transcriptase (Clontech, Mountain View, CA). TRA and TRB genes were amplified using pools of reverse primers using Taq polymerase (Invitrogen, Mulgrave, Victoria, Australia), as described by Wang et al. (27). PCR product was sequenced and the T-cell receptor (TCR) genes identified by alignment with the IMGT database (www.imgt.org/IMGT_vquest/vquest) (28).

Genomic DNA was purified from five million cloned T cells using QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany). Genotyping of the single nucleotide polymorphism, rs689, was performed using an inventoried TaqMan single nucleotide polymorphism genotyping assay obtained from Applied Biosystems (Life Technologies, Carlsbad, CA). The assay was run on a Roche LightCycler 480 II using Roche Probe Master Mix (Roche Applied Science, Penzberg, Germany) according to the manufacturer’s instructions.

The CFSE (5,6-carboxylfluorescein diacetate succinimidyl ester) proliferation assays were performed as described previously (15,22). Briefly, PBMCs were labeled with 0.1 μmol/L CFSE (Invitrogen, Grand Island, NY) and cultured (0.5 × 10⁶ cells in 0.5 mL) in sterile 5.0-mL tubes. CFSE-labeled PBMCs were cultured with either no antigen, proinsulin 40–52 peptide, proinsulin 48–62 peptide, or tetanus toxoid (10 LfU/mL). After 7 days of culture the cells were washed in PBS and stained on ice with anti-human CD4-Alexa Fluor 647 (clone OKT4, conjugated in house). Optimal compensation and gain settings for the flow cytometer for each experiment were determined based on unstained and single-stained samples. Propidium iodide was used to exclude dead cells. CD4⁺ T-cell proliferation was measured by determining the number of CD4⁺/CFSE^dim cells for every 5,000 CD4⁺/CFSE^bright cells. The results are presented as a cell division index (CDI), which is the ratio of the number of CD4⁺ cells that proliferated in the presence of antigen to the number that proliferated without antigen, as described previously (29).

To characterize human CD4⁺ T-cell responses associated with T1D, we obtained the pancreas from an organ donor who suffered from T1D. The organ donor was diagnosed with T1D at the age of 16 years. At diagnosis his blood glucose was 612 mg/dL (34 mmol/L; normal, 4.0–6.0 mmol/L) and C-peptide was 1.2 ng/mL (0.4 nmol/L). He had autoantibodies to GAD-65 and tyrosine phosphatase–related islet antigen 2. He was treated with short- (insulin glulisine) and long-acting insulin (insulin detemir). Death occurred from hypoglycemia 3 years after diagnosis, and organs were retrieved for donation after ∼24 h in critical care. Before organ donation his blood glucose was 144 mg/dL (8.0 mmol/L) and C-peptide was 0.3 ng/mL (0.1 nmol/L). His HLA type was HLA-A*01:01, 02:01; B*08:01, 51:01; DRB1*03:01, 04:04; DQA1*03:01, DQB1*03:02; DQA1*05:01, DQB1*02:01. Genetic analysis of the insulin promoter revealed that the donor was homozygous for the allele predisposing to T1D (class I VNTR).

Histological analysis of the donor’s pancreas revealed a wide range of islet appearance, including apparently normal, heavily infiltrated with residual insulin, and pseudoatrophic islets without insulin or infiltrate. Of the islets, 39% (143 of 371) had residual insulin staining; 70% (100 of 371) of insulin-positive islets were infiltrated by one or more T cells, and 25% of insulin-positive islets were infiltrated by >15 T cells (Fig. 1A). The infiltrate included CD4⁺ and CD8⁺ T cells (Fig. 1B), which was confirmed by flow cytometry (Fig. 1D). Thirty thousand islet equivalents were isolated to >90% purity. After overnight culture in CMRL 1066, the islets were cultured in RPMI medium/pooled human serum with IL-2 and IL-15. After 1 week cells were visible around 3–4 islets (Fig. 1C). These cells were CD3⁺/CD4⁺ and CD3⁺/CD8⁺ T cells (Fig. 1E), which were cloned and grown in vitro (21,26). No cells grew with islets isolated from two organ donors without T1D.

We isolated 62 CD4⁺ T-cell clones, of which 53 could be characterized in detail. Their antigen specificity was investigated by testing against a panel of synthetic peptides. All clones produced IFN-γ in response to phorbol 12-myristate 13-acetate/ionomycin stimulation, except four clones (1.5, 5.1, 4.16, and 5.8). An example of one clone (6.15) that responded to the pool of proinsulin peptides is shown (Fig. 1F–H). Of 53 clones, 14 (26.4%) responded to proinsulin peptides (Supplementary Fig. 1), but none of the clones responded to any of the other epitopes tested (data not shown). Clones that responded to the pool of 13 proinsulin peptides (Fig. 1F) were retested separately against each peptide. Ten clones, including 6.15, responded to proinsulin 37–54 (Fig. 1G). The remaining clones responded to proinsulin 43–60 or proinsulin 49–66. Fine epitope mapping (Fig. 1H) revealed that all 10 clones that responded to proinsulin 37–54 recognized the same core epitope (proinsulin 42–50, VELGGGPGA), but differences in their fine specificity were noted (Table 1 and Supplementary Fig. 2). Clone 4.13 responded to a slightly longer epitope (proinsulin 41–51, QVELGGGPGAG), and clone 5.5 responded to an epitope shifted one amino acid to the N-terminus (proinsulin 41–49, QVELGGGPG). Four clones specific for proinsulin 43–60 (1.1, 1.2, 2.4, and 3.15) recognized three distinct but overlapping epitopes: clone 1.1 recognized proinsulin 50–59 (AGSLQPLALE), 1.2 recognized proinsulin 50–58 (AGSLQPLAL), and clones 2.4 and 3.15 both recognized proinsulin 52–62 (SLQPLALEGSL) (Table 1 and Supplementary Fig. 2).

We next characterized the HLA restriction of the proinsulin-specific clones. The donor had the high T1D-risk HLA type: HLA-DR3,-DR4;-DQ2-DQ8. The responses of all 14 clones to proinsulin peptides were inhibited by a monoclonal antibody specific for HLA-DQ. Two examples, clone 6.15 (Fig. 2A) and clone 3.15 (Fig. 2C), are shown. (Other clones are shown in Supplementary Fig. 3.) The restricting HLA alleles were determined using a panel of cell lines transduced with individual HLA-DQ genes. Twelve clones responded to proinsulin peptide presented by HLA-DQ8 (DQA1*03:01, DQB1*03:02). Clone 6.15 is shown as an example (Fig. 2B). Clones 2.4 and 3.15 did not respond to HLA-DQ8 (DQA1*03:01, DQB1*03:02), but they responded to proinsulin peptides when presented by an HLA-DQ8 transdimer (DQA1*05:01; DQB1*03:02). Clone 3.15 is shown as an example in Fig. 2D. Data for all clones are shown in Supplementary Fig. 4. One clone (1.2) responded strongly to peptide presented by HLA-DQ8 but also responded to the same peptide presented by the HLA-DQ8 transdimer (DQA1*05:01; DQB1*03:02). Hence, all islet-infiltrating, proinsulin-specific CD4⁺ T-cell clones are restricted by HLA-DQ8 or the HLA-DQ8 transdimer.

To examine the clonal diversity of the islet-infiltrating CD4⁺ T cells, the TCR genes expressed by the clones were sequenced. The clones used a range of TRAV and TRBV genes (Fig. 3A and B). The most frequently expressed TRAV gene was TRAV26–1, which was expressed by 7 of 53 clones (13%). In contrast, usage of TRBV genes was more skewed; TRBV5–1*01 was expressed by 16 of the clones (30%). Translated TCR sequences for all clones are shown in Supplementary Table 3. No skewing of the CDR3 length was detected for TRAV and TRBV genes (Fig. 3C and D). In total, 47 unique clonotypes were expressed. We identified three sets of sister clones with identical TCRs (Fig. 3E). The first set comprised four clones: 1.9, 2.1, 3.14, and 4.6. These clones all used TRAV20*02 and TRBV5–1*01 and recognized proinsulin 42–50. The second set comprised three clones (5.8, 5.9, and 6.15) that used TRAV26–1*01 and TRBV5–1*01 and recognized proinsulin 42–50. The final set comprised two clones (2.4 and 3.15) that used TRAV19*01 and TRBV5–1*01 and responded to proinsulin 52–62. The remaining 44 clones expressed unique TRA/TRB combinations. The CDR3 DNA sequences of the proinsulin-specific TCRs are shown in Table 2.

Analysis of the clones’ antigen specificity showed that the clones isolated more than once all recognized proinsulin epitopes. Interestingly, the TCRβ chain, TRBV5–1*01, was used by four of five clone families specific for epitopes falling in proinsulin 40–51. Three of four TRVB5–1*01-expressing clones also expressed TRAV26–1*01. Collectively, the 14 clones specific for proinsulin C-peptide epitopes used eight distinct TRA/TRB combinations (Table 2). Although several clones used the same TRAV and TRBV, there was no evidence of conserved amino acids in CDR3 (Table 2 and Supplementary Table 3). This analysis revealed that several independent, proinsulin-specific clones infiltrated the islets of this donor.

Finally, we used a CFSE-based proliferation assay (29) to determine whether CD4⁺ T-cell responses to these defined epitopes were detectable in PBMCs from HLA-DQ8⁺ individuals with recent-onset (<3 months) T1D. Two of eight subjects had CD4⁺ T-cell responses (CDI ≥2.0) to proinsulin 40–52, and three of eight had responses to proinsulin 48–62 (Fig. 4A and B). No responses were detected from six HLA-DQ8⁺ subjects without T1D. CD4⁺ T-cell proliferation to tetanus toxoid was detected in all donors (Supplementary Tables 1 and 2). This concurs with previous reports using longer peptides, showing that CD4⁺ T-cell responses to this region of the C-peptide are detectable in the PBMCs of people with T1D (16).

We made the following discoveries from our analysis of the islet-infiltrating CD4⁺ T cells from the pancreas of a deceased organ donor who had T1D: 1) CD4⁺ T-cell responses were directed against six distinct but overlapping epitopes from the C-peptide of proinsulin; 2) all proinsulin-specific islet-infiltrating CD4⁺ T-cell clones recognized peptides presented by HLA-DQ; 3) a pair of clones responded to a proinsulin epitope presented by the HLA-DQ8 transdimer; and 4) most (five of eight) proinsulin-specific clones expressed TRBV5–1*01.

To our knowledge, this is the first report to reveal that HLA-DQ8-restricted, proinsulin-specific CD4⁺ T cells infiltrated the pancreatic islets of a patients with T1D. More than one in four of the clones were specific for proinsulin, but we did not detect any clones specific for insulin or any clones specific for any of the other 26 epitopes tested. Our results suggest that HLA-DQ-restricted CD4⁺ T cells specific for proinsulin are abundant within the islets and central to the pathogenesis of T1D. The absence of insulin-specific T-cell clones is consistent with reports that HLA-DQ-restricted regulatory T-cell responses toward insulin are detectable in individuals with T1D and their unaffected siblings (30).

The donor’s homozygous type 1 insulin VNTR genotype may have promoted the broad T-cell response against proinsulin. HLA-DR4-restricted CD4⁺ T cells specific for proinsulin 52–66 were most frequently detected in the blood of subjects who were VNTR I homozygous (14). Our analysis of the TCR genes expressed by human islet-infiltrating CD4⁺ T cells revealed that the 53 clones expressed 47 unique clonotypes. We cannot exclude the possibility that the isolation of multiple clones with unique TCRs arose from skewing of the repertoire of infiltrating T cells occurred while the islets were cultured in the presence of cytokines, rather than in vivo before islet isolation. Nonetheless, eight unique proinsulin-specific clonotypes were present, arguing against the possibility that proinsulin-specific CD4⁺ T cells were grossly favored during the isolation.

The T cells we isolated were not selected based on their antigen specificity, yet the epitopes recognized fall in a narrow region of the C-peptide of proinsulin. Several clones recognize epitopes containing the sequence—GGGPGA (see Table 2)—which is rich in amino acids with very small side chains. It remains to be determined whether this very “bland” epitope is recognized by CD4⁺ T cells with degenerate specificity, as has been reported for proinsulin-specific CD8⁺ T cells (31). This degenerate specificity may allow these proinsulin-reactive T cells to escape thymic deletion yet become activated in the islets, where the concentration of proinsulin is very high. The antigen specificity of many of the clones was not defined. Most of these clones may respond to other as yet undefined β-cell antigens. It remains possible that the four clones that did not make IFN-γ respond to some of the peptides tested here but were not detected because they secrete cytokines other than IFN-γ.

There have been few previous reports of HLA-DQ8-restricted responses against β-cell antigens (32). All 14 T-cell clones reported in this study recognized proinsulin peptides presented by HLA-DQ8 or HLA-DQ8 transdimer. There was no evidence of a unique HLA-DQ8-transdimer–restricted epitope; the HLA-DQ8-transdimer–restricted epitope overlaps almost completely with two HLA-DQ8-restricted epitopes. Hence, these data support the hypothesis that HLA-DQ2; DQ8 heterozygous individuals are at the greatest genetic risk of T1D because they have a relatively high density of T1D-associated HLA molecules on the surface of their APCs.

Pancreatic sections from the donor studied here were included in the analysis of human islet-infiltrating CD8⁺ T cells by Coppieters et al. (33). This analysis revealed that HLA-A2-restricted CD8⁺ T cells specific for islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP), GAD-65, preproislet amyloid polypeptide, and preproinsulin were present in this donor’s islets. It is believed that HLA class I–restricted CD8⁺ T cells that recognize β-cell autoantigens are responsible for β-cell destruction (34). In the NOD mouse model, spreading of the CD8⁺ immune response from proinsulin to IGRP only requires CD4⁺ T cells specific for proinsulin (11), which could also be the case in humans. Identification of HLA-DQ8-restricted epitopes strongly implicated in β-cell autoimmunity will facilitate the analysis of CD4⁺ T-cell responses in T1D.

We did not attempt to characterize the memory/activation status of the islet-infiltrating CD4⁺ T cells. To obtain sufficient cells to characterize, we stimulated the T cells with anti-CD3 in the presence of cytokines. This treatment activates the cells and drives their proliferation, precluding an analysis of their activation status. We expect these cells would have an effector memory phenotype in vivo, consistent with their role in mediating β-cell destruction. This issue could be addressed using HLA-DQ8/C-peptide tetramers to stain proinsulin-specific CD4⁺ T cells, combined with antibody staining for activation/memory markers. However, this analysis will be limited by the small number of cells that can be recovered from human islets.

Here we report an analysis of islet-infiltrating CD4⁺ T cells from a single donor. Further work will be required to determine the variability of islet-infiltrating CD4⁺ T cells in different individuals. Analysis of T-cell responses in PBMCs underscores the variability inherent in human studies. This presumably reflects differences in the specificity of pathogenic CD4⁺ T-cell response between individuals. The capacity to detect CD4⁺ T-cell responses may allow antigen therapy to be targeted toward individuals who have an existing response to these epitopes.

Antigen-specific therapies are a very attractive approach to treating T1D. Toward this goal, an 18mer proinsulin peptide (proinsulin 51–68), encompassing the epitopes recognized by clones 1.1, 1.2, 2.4, and 3.15, has been tested in a phase 1 human trial (35). Our work, showing that proinsulin-specific CD4⁺ T cells are a major component of the islet CD4⁺ T-cell infiltrate, adds further support to the notion that CD4⁺ T cells against proinsulin are pathogenic and therefore are attractive candidates for use in antigen-specific therapy protocols.

Acknowledgments. The authors thank Dr. Zhenjun Chen (University of Melbourne) for advice and assistance in the generation of retrovirally transduced cell lines.

Funding. S.I.M., V.P., and J.P.K. have received JDRF grant 17-2011-527. A.W.P. and N.L.D. have received JDRF project grant 17-2012-134. S.I.M. and A.W.P. have received National Health and Medical Research Council (NHMRC) grant APP1061961. S.I.M. and T.W.H.K. have received support from the Diabetes Australia Research Trust. S.I.M., T.W.H.K., H.E.T., B.K., V.P., and J.P.K. have received support from the Operational Infrastructure Support Program of the Victorian Government. T.W.H.K. and H.E.T. have received NHMRC program grant APP1037321. T.W.H.K., H.E.T., and T.L. have received JDRF/Department of Health and Aging grants, and P.T.C.H. has received a grant for islet transplantation. K.K. has received NHMRC project grant AI1008854. H.E.T. is supported by a fellowship from the NHMRC (APP1042735). K.K. is the recipient of an NHMRC CDF2 Fellowship. P.J.O. is an NHMRC senior practitioner fellow. A.W.P. is an NHMRC Senior Research Fellow.

Duality of Interest. No conflicts of interest relevant to this article were reported.

Author Contributions. V.P., J.P.K., P.D.C., T.C.B., and S.I.M. performed experiments. B.K., T.L., P.T.H.C., P.J.O., C.R., P.B., E.H., H.E.T., and T.W.H.K. provided clinical samples. T.C.B., K.K., A.W.P., N.L.D., H.E.T., T.W.H.K., and S.I.M. provided critical reagents and intellectual input into the planning and execution of the work. T.W.H.K. and S.I.M. wrote the manuscript. S.I.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Part of this work was presented at the Australian Diabetes Society Annual Scientific Meeting, Sydney, Australia, 28–30 August 2013; at the Immunology of Diabetes Society 13th International Congress, Lorne, Victoria, Australia, 7–13 December 2013; and at the Federation of Clinical Immunology Societies Meeting, Chicago, IL, 25–28 June 2014.