GAD-alum immunotherapy in type 1 diabetes expands bifunctional Th1/Th2 autoreactive CD4 T cells

Abstract

Aims/hypothesis: Antigen-specific therapy aims to modify inflammatory T cell responses in type 1 diabetes and restore immune tolerance. One strategy employs GAD65 conjugated to aluminium hydroxide (GAD-alum) to take advantage of the T helper (Th)2-biasing adjuvant properties of alum and thereby regulate pathological Th1 autoimmunity. We explored the cellular and molecular mechanism of GAD-alum action in the setting of a previously reported randomised placebo-controlled clinical trial conducted by Type 1 Diabetes TrialNet.

Methods: In the clinical trial conducted by Type 1 Diabetes TrialNet, participants were immunised with 20 μg GAD-alum (twice or three times) or alum alone and peripheral blood mononuclear cell samples were banked at baseline and post treatment. In the present study, GAD-specific T cell responses were measured in these samples and GAD-specific T cell lines and clones were generated, which were then further characterised.

Results: At day 91 post immunisation, we detected GAD-specific IL-13⁺ CD4 T cell responses significantly more frequently in participants immunised with GAD-alum (71% and 94% treated twice or three times, respectively) compared with those immunised with alum alone (38%; p = 0.003 and p = 0.0002, respectively) accompanied by high secreted levels of IL-13, IL-4 and IL-5, confirming a GAD-specific, GAD-alum-induced Th2 response. Of note, GAD-specific, IL-13⁺ CD4 T cells observed after immunisation co-secreted IFN-γ, displaying a bifunctional Th1/Th2 phenotype. Single-cell transcriptome analysis identified IL13 and IFNG expression in concert with the canonical Th2 and Th1 transcription factor genes GATA3 and TBX21, respectively. T cell receptor β-chain (TCRB) CDR3 regions of GAD-specific bifunctional T cells were identified in circulating naive and central memory CD4 T cell pools of non-immunised participants with new-onset type 1 diabetes and healthy individuals, suggesting the potential for bifunctional responses to be generated de novo by GAD-alum immunisation or via expansion from an existing public repertoire.

Conclusions/interpretation: GAD-alum immunisation activates and propagates GAD-specific CD4 T cells with a distinctive bifunctional phenotype, the functional analysis of which might be important in understanding therapeutic responses.

Keywords: Autoreactive; Epitopes; GAD; GAD-alum; Glutamic acid decarboxylase; IL-13; Immunotherapy; T cell receptor; T cells; TCR; Th2; TrialNet.

Fig. 1

Increase in GAD-specific IL-13 responses in GAD-alum-treated participants. (a) Representative ELISpot images from a GAD-alum-treated patient at baseline and at day 91 post treatment. Spots represent IL-13 production by single CD4 T cells. IL-13 responses are unusual in type 1 diabetes but are induced after GAD-alum immunotherapy, as demonstrated in (b) showing increased frequency of GAD-specific IL-13 responses post vaccination with GAD-alum (twice [GAD×2] or three times [GAD×3], but not with alum alone (IL-13 stimulation index [SI] at day 0 [D0; baseline, first treatment] and day 91 [D91]). SI represents the number of spots after stimulation in vitro with recombinant human GAD (rhGAD65) divided by the number of spots after control stimulation. Differences between D0 and D91 for each of treatment group are shown using paired statistical comparisons using Mann–Whitney U tests (**p < 0.01; ***p < 0.001). (c) Responses can also be represented as fold change from baseline in participants receiving alum alone (circles) or GAD-alum twice (GAD×2) or three times (GAD×3). Statistical comparison of fold change is shown (***p < 0.001). (d, e) Induced GAD-specific IL-13 responses (shown as SI at day 91) correlate with change in anti-GAD autoantibodies (GADA) between baseline and day 91 in GAD-alum-treated participants (d) but not in those given alum alone (e). Note that post treatment GADA data are not available for the participant in the GAD×3 group with SI = 289 in (b) and (c). GADA levels were expressed as index units (antibody index value relative to standard serum)

Fig. 2

Immunogenic regions of GAD65. Regions were mapped by screening overlapping GAD65 peptides with GAD-specific T cell lines generated using PBMCs from participants immunised with GAD-alum and tested for IL-13 response by ELISpot. Lines were generated from participants homozygous for HLA-DR3/DQ2 (DRB1*0301/DQA1*0501-DQB1*0201), homozygous for HLA-DR4/DQ8 (DRB1*0404/DQA1*0301 DQB1*0302) and heterozygous for HLA-DR3/4 DQ2/8. Epitope regions are shown in red when they are observed in only one line, orange when observed in two lines and yellow when multiple lines show reactivity

Fig. 3

GAD epitopes identified by screening overlapping peptides for IL-13 responses

Fig. 4

Examples of GAD-specific bifunctional Th1/Th2 cells induced upon GAD-alum immunisation. Representative data from two GAD-alum-vaccinated participants showing responses to GAD65 peptides (a) 231–250 and (b) 556–575 at baseline and day 84 (D84) post immunisation analysed by dual IL-13/IFN-γ indirect FluoroSpot. In the merged image panels, dual-cytokine-secreting cells appear yellow. (c) Baseline and day 84 responses to the recall antigen Pediacel for comparison. IFN-γ responses are shown in green, IL-13 in red and bifunctional Th1/Th2 responses are yellow

Fig. 5

GAD-specific bifunctional cells are induced upon GAD-alum immunisation. IFN-γ and IL-13 responses to GAD epitopes: GAD226-245 (a, e, i), GAD231-250 (b, f, j), GAD556-575 (c, g, k) and Pediacel (d, h, l) were analysed by dual IL-13/IFN-γ indirect FluoroSpot in nine participants immunised with GAD-alum at baseline (B/L) and at day 84 (D84). Bifunctional Th1/Th2 responses are shown in (a–d), single IL-13 in (e–f) and IFN-γ responses in (i–l). Each symbol represents a different immunised participant. The comparison between responses at baseline and D84 post immunisation were analysed by Wilcoxon matched-pairs signed rank tests (*p < 0.05)

Fig. 6

GAD-specific CD4 T cells induced by GAD-alum immunisation are bifunctional. (a) Representative GAD-specific CD4 T cell clone stimulated with diluent DMSO or (b) PMA/ionomycin for 3 h and stained for intracellular expression of IL-13 and IFN-γ shows the clone’s bifunctional potential. (c) GAD peptide-specific clone exhibits a bifunctionality when stimulated with cognate GAD65 555–567 peptide in vitro and analysed by dual-cytokine FluoroSpot. The control condition is stimulation with DMSO (the peptide diluent). The Th1/Th2 phenotype (IFN-γ⁺/IL-13⁺) is shown in yellow. IFN-γ responses are green and IL-13 are red (representative data from six clones). (d) Expression of cytokine and transcription factor mRNA by single sorted clone/line cells detected by PCR shows that a majority of GAD-alum induced bifunctional CD4 T cells co-express Th1 and Th2 differentiation markers TBX21 (encoding T-bet), IFNG and GATA3 as well as IL4 and/or IL13

Fig. 7

T cell receptors of Th1/Th2 bifunctional cells. Th1/Th2 bifunctional cells induced upon GAD-alum immunisation have the potential to arise after priming from naive cells and/or expansion from GAD-specific central memory CD4 T cells. TCRB CDR3s from bifunctional cells from GAD-specific T cell lines generated post GAD-alum immunisation were sequenced; these were then compared against TCRB CDR3s bulk sequenced from naive and central memory CD4⁺ T cell subsets from participants with type 1 diabetes (n = 14) and healthy controls (n = 17) (true naive, TN; central memory, CM) [18]. The CDR3s from bifunctional Th1/Th2 cells are identifiable in both (a) naive and (b) central memory cells. The probability of generation (pGEN) values are shown on the bottom of the panels: values ranging from 10⁻⁷ to 10⁻¹⁰ indicate a high chance of a TCR being generated and therefore likely to be public. (c) The prevalence of clonotypes amongst all participants