Regulatory T cells induced by single peptide liposome immunotherapy suppress islet-specific T cell responses to multiple antigens and protect from autoimmune diabetes

Abstract

Antigen-specific tolerizing immunotherapy is considered the optimal strategy to control type 1 diabetes (T1D), a childhood disease involving autoimmunity towards multiple islet antigenic peptides. To understand whether tolerizing immunotherapy with a single peptide could control diabetes driven by multiple antigens, we co-encapsulated the high-affinity CD4⁺ mimetope (BDC2.5_mim) of islet autoantigen Chromogranin A (ChgA) with or without calcitriol (1α,25-dihydroxyvitamin D3) into liposomes. After liposome administration, we followed the endogenous ChgA-specific immune response with specific tetramers. Subcutaneous but not intravenous liposome administration induced ChgA-specific Foxp3⁺ and Foxp3⁻ PD1⁺ CD73⁺ ICOS⁺ IL-10⁺ pTreg in pre-diabetic mice, and liposome administration at the onset of hyperglycemia significantly delayed diabetes progression. After BDC2.5_mim/calcitriol liposome administration, adoptive transfer of CD4⁺ T cells suppressed the development of diabetes in NOD-SCID mice receiving diabetogenic splenocytes. After BDC2.5_mim/calcitriol liposome treatment and expansion of ChgA-specific pTreg, IFN-γ production and expansion islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP)-specific CD8⁺ T cells was also suppressed in pancreatic draining lymph node, demonstrating cross-tolerance at the site of antigen-presentation. Thus, liposomes encapsulating the single CD4⁺ peptide, BDC2.5_mim and calcitriol induce ChgA-specific CD4+ T cells which regulate CD4⁺ and CD8⁺ self-antigen specificities and autoimmune diabetes in NOD mice.

Introduction

Type 1 diabetes (T1D) is an autoimmune disease of the pancreatic islets in which antigen-specific T cells destroy islet beta cells, resulting in loss of insulin production and hyperglycemia. The incidence of T1D in young children and adolescents has increased by about 4% per year in recent decades [1]. Despite a good understanding of the immunology of T1D, none of the >70 phase 1–3 clinical trials of drugs targeting immune effector pathways, immune cell subsets or inflammation in the last 10 years has achieved substantive benefits for patients [2]. Nonetheless, anti-CD3 (Teplizumab), which delayed disease progression in a subset of recent-onset subjects and in high-risk pre-diabetic subjects [3], expanded PD-1⁺ Foxp3⁺ regulatory T cells (Treg) and restrained CD8 T cell activation [4], suggesting that approaches enhancing tolerance may ultimately prove effective for T1D control.

Development of antigen-specific immunotherapy is a major goal for T1D, as it should improve specificity for the autoreactive process, enhancing tolerability and long-term immunomodulatory effects. To this end, several studies have demonstrated antigen-specific tolerance in NOD mice with various antigen preparations, including s.c. islet antigenic peptide infusion, plasmid-encoded peptide, peptide-I-A^g7-coated nanoparticles [5], peptide-coupled apoptotic splenocytes [6] or nanoparticles [7] and nanoparticles encapsulating islet protein with ITE [8]. Islet peptides [9, 10] including chromogranin A (ChgA) mimetope (BDC2.5_mim) [5, 11–13] or 2.5-hybrid peptide [7] have been used to confer islet specificity. Trials using a single soluble peptide or plasmid-encoded peptide [14, 15] demonstrate safety, tolerability and some signs of immune modulation.

Delivery of free islet-antigen to individuals with autoimmunity runs the risk of activating effector T cells rather than inducing tolerance [2]. Dendritic cells (DCs) in both NOD mice and humans with T1D were shown to have constitutive NF-κB activation and an activated phenotype [16, 17], which may pose a barrier to tolerance induction. Previously, we showed that liposomes could be used to deliver an NF-κB inhibitor and antigen to APCs, resulting in Treg induction and suppression of inflammatory arthritis [18, 19]. In this current study, we co-encapsulated BDC2.5_mim and the NF-κB inhibitor, 1α,25-dihydroxyvitamin D3 (calcitriol) in liposomes then used I-A^g7/BDC2.5_mim tetramers to elucidate their impact on endogenous ChgA-specific CD4⁺ T cells in the physiological setting of the NOD mouse model of disease.

Materials and Methods

Animals

Female non-obese diabetic (NOD/ShiLtJArc, referred to as NOD) and NOD.SCID mice were obtained from the Animal Resources Centre (Cunning Vale, Australia). NOD-8.3 TCR transgenic (CD8⁺, IGRP-specific) [20] and NOD-BDC2.5 TCR transgenic (CD4⁺, chromogranin A specific) [21] were purchased from Jackson Laboratories and were intercrossed with NOD-Thy1.1⁺ mice [22]. Animals were kept at the pathogen free Biological Resources Facility within the Translational Research Institute (Brisbane, Australia). Female, non-diabetic mice, aged as noted below were used in all experiments. All experiments were approved by the University of Queensland animal ethics committee.

Liposome Formulation and Treatment

Liposomes were prepared by the thin film hydration method [23]. Briefly, L-α-phosphatidylcholine (egg) (EPC; Avanti Polar Lipids, Alabaster, AL) and L-α-phosphatidylglycerol (egg) (EPG, Avanti Polar Lipids) were dissolved in a 9:1 ratio in chloroform/ethanol and evaporated to prepare a thin film under reduced pressure in a rotary evaporator. The film was hydrated with HEPES buffer (pH 7.4) and the resulting liposomes were subjected to several freeze-thaw cycles followed by sonication using a 3 mm probe at 40% amplitude on ice (Sonics and Materials, Newton, CT). For the preparation of drug loaded liposomes, 400ng/mL calcitriol (Dishman, Veendaal, Netherlands) was added to the lipid film phase while 30μg/mL peptide (BDC2.5_mim, AHHPIWARMDA [24], HEL_11–25 (AMKRHGLDNYRGYSL [25]) (GLBiochem, Shanghai, China) were added to the aqueous phase during hydration. The liposome characteristics in vitro and their in vivo behavior were comparable to those previously published for OVA_323–339/calcitriol liposomes [26]. Liposomes were administered i.v. (100 μL) or s.c. (50 μL in each flank) to 8–10 week old NOD mice. In most experiments, mice received 2 liposome injections at day 0 and day 7, and pancreatic lymph nodes (PLNs) and spleens were harvested at day 11.

In vitro DC:T co-cultures

NOD CD11c⁺ DC were magnetically sorted using CD11c MicroBeads UltraPure from Miltenyi Biotec and loaded for 2hrs at 37 degrees with BDC2.5 mimotope or varying concentrations of BDC2.5_mim/D3 or BDC2.5_mim or blank liposomes. Cells were washed twice and 5 × 10⁴ DCs were cultured in vitro with 1 × 10⁵ CD4⁺ CD90.1⁺ NOD BDC2.5 transgenic T cells, magnetically sorted using CD4⁺ T cell isolation kit from Miltenyi Biotec. Some T cells were labelled with CellTrace™ Violet Cell Proliferation kit as per instruction (ThermoFischer Scientific). Cells cultured for 18 to 72hrs and then washed and labeled with TCRβ, CD4, CD69, CD90.1 and CD44 mAb for FACS analysis. Some cells were cultured in the presence of anti-MHC class II I-A^g7 blocking antibody or isotype [27].

Survival experiments and assessment of diabetes

In prevention experiments, 8-week old female NOD mice received 4 injections s.c. or i.v. of 100μl liposomes on days 0, 7, 14 and 28. Control mice received PBS or HEL/D3 as indicated. Blood glucose levels were measured twice a week in female NOD mice with a Freestyle Optium Blood Glucose Monitoring device (Abbott) starting at 12 weeks. Mice with persistent blood glucose reads >360 mg/dL were designated as diabetic and culled. Blood was collected weekly by retro-orbital eye bleed and red blood cells were lysed by adding ammonium-chloride-potassium buffer before FACS analysis.

For therapeutic experiments, NOD mice were monitored twice weekly and considered hyperglycaemic after two consecutive blood glucose concentrations >198 mg/dL. Mice were then randomized to treatment groups and injected s.c. or i.v. twice weekly for 4 weeks with PBS or liposomes. Mice with blood glucose reads >288 mg/dL were treated with insulin (0.75U/kg) daily for one week to control glycosuria after diabetes onset [5]. Insulin was then withdrawn and blood glucose recorded every 2–3 days. Mice were culled when persistently recording blood glucose level >360 mg/dL and/or required culling for ethical reasons.

In the accelerated model of diabetes, 8-week old donor NOD mice were treated s.c. or i.v. with 100μl of BDC2.5_mim/D3 liposomes or PBS at day 0 and day 7. Four days later, mice were culled and 5 × 10⁶ splenocytes or 1.5 × 10⁶ sorted CD4⁺ or 3.5 × 10⁶ CD4⁻ cells were transferred to 8–10 week old NOD.SCID, mice together with 10 × 10⁶ spleen cells harvested from diabetic NOD mice. For IL-10R blocking, mice received 4 bi-weekly x 500μg of antibody (clone 1B1.3A, BioXcell) or isotype control (clone HRPN, BioXcell) at day 5–20 or day 15–28. Blood glucose levels of NOD.SCID mice were then monitored and mice were culled when persistently recording blood glucose level >360 mg/dL and/or required culling for ethical reasons.

Cytokine production and Bystander experiment

Female NOD mice received 2 s.c. injections of PBS or 100μl BDC2.5_mim/D3 liposomes on days 0 and 7. Prior to cytokine analysis, mice were challenged with 10μg peptide in PolyI:C (100μg)/IFA (100μl) (Sigma-Aldrich, USA) emulsion on day 16 or 20 and analyzed 5 days later. Inguinal (iLN), pancreatic (PLN) and spleen were collected and cells were restimulated in vivo using PMA (10ng/ml)/ionomycin (500ng/ml) for 3.5hrs in the presence of Brefeldin A (2.5μg/ml). For bystander experiments, mice received 1 × 10⁵ CD90.1⁺ IGRP-specific 8.3 CD8⁺ T cells on day 14 after liposome injections at day 0, 7 and 14 and were analyzed 5 days later using CD90.1 marker. Endogenous IGRP-specific T cells were analyzed at day 10 after 2 liposome injections on day 0 and 7 and traced using I-A^g7-IGRP tetramers.

Flow cytometry

For the in vitro experiment, cells were then labelled using the following antibodies in FACS buffer for 30 min in the dark on ice: anti-CD4 APC-Cy7 (GK1.5), anti-CD69 Pacific Blue (H1.2F3), anti-TCRb Alexa Fluor 488 (H57–597), anti-CD90.1 PerCP/Cy5.5 (OX-7), anti-CD44 APC (IM7), anti-CD11c BV605 (N418), anti-CD11b Alexa Fluor 700 (M1/70), anti-class I-A^d biotin (39–10-8) and Streptavidin PE-Cy7 (Biolegend). Live/Dead stain Aqua (Thermofisher Scientific) was used to exclude dead cells.

Peptide loaded monomers were provided by the NIH Tetramer Core Facility (Georgia, USA) and tetramers were conjugated using RPE-streptavidin (ProZyme) according to instructions provided by the facility. CLIP specific tetramers were used as negative controls for the MHC-II tetramer staining. Cells were first incubated for 20–30 min at room temperature with tetramers only. I-A^g7-BDC2.5_mim tetramers were used to stain ChgA-specific T cells or H-2K^d-IGRP_206–214 tetramers were used to stain IGRP-specific T cells. Cells were then labelled using the following antibodies in FACS buffer for 30 min in the dark on ice: anti-CD4 APC-Cy7 (GK1.5), anti-CD73 Pacific Blue (TY/11.8), anti-CD279/PD-1 BV605 (29F.1A12), anti-CD278/ICOS FITC (C398.4A), anti-CD44 PE-Cy7 (IM7), anti-CD62L BV421 (clone MEL-14) (Biolegend). Live/Dead stain Aqua (Thermofisher Scientific) was used to exclude dead cells. Foxp3 Alexa-700 (FJK-16s) was stained intracellularly using the eBioscience Foxp3 staining kit. In some experiments, cells were restimulated in vitro with PMA and ionomycin in the presence of Brefeldin A for 3.5hrs, then intracellularly labelled with Foxp3 (eBioscience), IL-10 PE (clone JES5) and IFNg APC (clone XMG1.2) from Biolegend. After washing, 20μl of fluorobeads (Beckman Coulter Flow-count Fluorospheres) were added to each sample. Cells were acquired using a BD LSRFortessa X-20. All flow cytometry data were analyzed using FlowJo (version X10) software.

Statistical Analysis

Survival curves were analyzed using a Mantel-Cox log rank test. A one-way ANOVA was used to analyze other data. Linear regression test was used for correlation. All statistical analyses were performed using GraphPad Prism 8 software, with a significance threshold of p ≤ 0.05. *p≤ 0.05, **p≤ 0.01, ***p≤ 0.001 and ****p≤ 0.0001

Results

BDC2.5 T cells activation using liposomes is MHC-II dependent and expands Foxp3⁺ T cells.

To test the response to liposomal co-delivery of a high affinity auto-antigenic peptide and a NF-κB inhibitor, we encapsulated the BDC2.5_mim that is recognized by endogenous BDC2.5 CD4⁺ T cells specific for the islet derived antigen ChgA [13, 21], into liposome nanoparticles with or without calcitriol (D3). CD11c⁺ DC loaded in vitro with BDC2.5_mim/D3 liposomes but not blank liposomes activated transgenic T cells in a dose dependent manner (Figure 1A) and T cell activation was inhibited by anti-MHC class II mAb (Figure 1B). The proliferation of CD44^hi BDC2.5 transgenic T_EM cells stimulated in the presence of BDC_mim and BDC2.5_mim/D3 liposomes, with no reduction in the presence of calcitriol (Figure 1C). After a single administration of liposomes to NOD mice transferred with ChgA-specific T cells, ChgA-specific T cells expanded in spleen and PLN at day 4 and contracted at day 11, with a concomitant expansion of Foxp3⁺ T cells (Figures 1D and E). While antigen-specific T cells expanded to a greater extent in spleen and pLN in response to 2.5_mim/D3 than 2.5_mim liposomes at day 4, antigen-specific T cells contracted and Treg expanded by day 11 in response to each liposome at each site. These data indicate that BDC2.5_mim and BDC2.5_mim/D3 liposomes drive an MHC class II-dependent regulatory response in TCR-transgenic T cells in vitro and in vivo.

Figure 1. BDC2.5 T cell activation using liposomes is MHC class II-dependent and expands Foxp3⁺ T cells.

A-C) NOD CD11c⁺ DC were MACS sorted and loaded for 2hrs at 37 degrees with BDC2.5 mimotope or various dose of BDC2.5mim/D3 or blank liposomes. Cells were washed and 5 × 10⁴ CD11c⁺ DCs were cultured in vitro with 1 × 10⁵ CD4⁺ BDC2.5 transgenic MACS sorted T cells. A) Percentage of CD69 expression on live TCRβ⁺ CD4⁺ T cells and B) percentage of inhibition of CD69 upregulation in the presence of varying concentrations of anti-MHC class II blocking mAb after 18hrs. C) Percentage proliferation of TCRβ⁺CD90.1⁺CD44^hiCD4⁺ T cells. T cells alone served as negative controls and T cells stimulated with anti-CD3 mAb served as positive controls. N =3–6, 2 experiments. D, E) 8–10 week old female NOD mice received 10⁶ CD90.1⁺ BDC2.5 transgenic cells at day 0. Mice were analyzed at day 4 and 11 after one i.v. injection of 100μl liposomes encapsulating BDC2.5_mim/D3 or BDC2.5_mim. Controls received PBS. D) Percentages of transferred ChgA-specific T cells identified with a CD90.1 congenic marker and I-A^g7-ChgA-specific CD4 tetramer in live CD3⁺ CD4⁺ T cells in mice. E) Proportion of CD90.1⁺ ChgA-specific Foxp3⁺ cells in spleen (left) and pLN (right). n= 3–5 from 2 experiments. One-way ANOVA with Tukey’s multiple comparisons test. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

BDC2.5_mim/D3 liposomes expand endogenous Foxp3⁺ ChgA-specific CD4⁺ T cells.

To determine the impact on endogenous ChgA-specific T cells, liposomes were administered s.c. or i.v. into NOD mice on days 0 and 7 and endogenous ChgA specific CD4⁺ T cells were analyzed using I-A^g7/BDC2.5_mim tetramers on day 11. The proportion of tetramer⁺ endogenous ChgA-specific CD4⁺ T cells increased in spleen but not PLN after administration of s.c. and i.v. liposomes encapsulating BDC2.5_mim with or without calcitriol, compared to liposomes encapsulating HEL/D3 or empty liposomes (Figure 2A–C). The proportion of endogenous ChgA-specific Foxp3⁺ CD4⁺ T cells increased in spleen (Figure 2D) after s.c. but not i.v. liposomes. This splenic Treg expansion was reflected in absolute cell counts (Figure S1). ChgA-specific Treg did not expand in PLN (Figure 2C, E).

Figure 2. Subcutaneous but not intravenous BDC2.5_mim/D3 liposome treatment expands endogenous Foxp3⁺ ChgA-specific CD4⁺ T cells.

Female NOD mice (8–10 weeks old) were analyzed at day 11 after two (day 0, 7) s.c. or i.v. injections of 100μl liposomes encapsulating BDC2.5_mim with or without D3. Control mice received PBS, HEL/D3 or empty liposomes. A) Representative dot plots of endogenous ChgA-specific cells identified using I-A^g7-BDC2.5_mim CD4 tetramers within live CD3⁺ CD4⁺ T cells. B, C) Percentage of ChgA-specific endogenous T cells in spleen (B) and PLN (C). D, E) Proportion of Foxp3⁺ T cells within ChgA-specific T cells in spleen (D) and PLN (E). n = 3–16 from 3–4 experiments, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, one-way ANOVA

In the same experiment, we observed an increase in both the tetramer and Foxp3 MFI in spleen and PLN (Figure 3A,B) in response to liposomes encapsulating BDC2.5_mim relative to PBS, consistent with increased TCR affinity and an enhanced regulatory phenotype of the antigen-specific T cells. No variation was observed in non-ChgA-specific T cells in spleen or PLN (Figure 3C). Tetramer MFI and Foxp3 MFI increases were correlated in spleen after s.c. or i.v. liposome treatment (Figure 3D). ICOS intensity also strongly correlated with Foxp3 MFI after s.c. but not i.v. liposome treatment in spleen, consistent with induction of an effector Treg phenotype (Figure 3E). These data indicate that liposomes encapsulating BDC2.5_mim with or without calcitriol increase the proportion of endogenous antigen-specific Foxp3⁺ Treg cells in lymphoid organs after s.c. delivery.

Figure 3. Foxp3 MFI correlates with tetramer MFI and ICOS MFI following s.c. injections of liposomes.

Same experiment as Figure 1. A) I-A^g7-BDC2.5_mim tetramer staining intensity (MFI) or B) Foxp3 MFI of endogenous ChgA-specific T cells 11 days after each treatment (fold increase) relative to PBS controls in spleen and PLN. C) Foxp3 MFI fold increase relative to PBS for I-A^g7-BDC2.5_mim tetramer-negative CD4⁺ T cells. D) Correlation between Foxp3 MFI and tetramer MFI fold increase relative to PBS for ChgA-specific CD4⁺ T cells. E) Correlation between Foxp3 MFI and ICOS MFI fold increase relative to PBS for Foxp3⁺ ChgA-specific CD4⁺ T cells. n = 3–16 from 3–4 experiments, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, one-way ANOVA (A-C), linear regression (D,E).

Foxp3⁻ ChgA-specific CD4⁺ endogenous T cells are antigen-experienced and produce IL-10.

While liposomes encapsulating BDC2.5_mim increased the proportion of splenic antigen-specific Foxp3⁺ Treg, the majority of antigen-specific T cells remained Foxp3⁻. Relative to PBS treatment, ChgA-specific Foxp3⁻ CD4⁺ T cells expanded by BDC2.5_mim/calcitriol liposomes displayed evidence of antigen-experience, with an effector memory (CD62L^lowCF44^high, T_EM) phenotype in spleen (Figure 4A), and co-expression of high levels of ICOS (Figure 4B), PD1 and CD73 (Figure 4C) in spleen and PLN. ICOS, PD1 and CD73 levels were similar to those expressed by Foxp3⁺ ChgA-specific CD4⁺ T cells. A similar T_EM phenotype with co-expression of ICOS, PD1 and CD73 by antigen-specific Foxp3⁻ T cells occurred after i.v. administration of liposomes encapsulating BDC2.5_mim (Figure S2). To assess the cytokine production of ChgA-specific CD4⁺ T cells, liposome-treated mice were challenged with BDC2.5_mim peptide emulsified with IFA and polyI:C, and harvested cells were restimulated ex vivo with PMA/ionomycin 5 days later. In both spleen and PLN, ChgA-specific Foxp3⁻ IL-10⁺ cells were 2.5 to 5 times more frequent (Figure 4D, F) and splenic ChgA-specific Foxp3⁻ IFNγ⁺ cells decreased by 2-fold after liposome relative to PBS administration (Figure 4E, G). Whereas ChgA-specific Foxp3⁺ IL-10⁺ cells did not differ with liposome treatment, fewer ChgA-specific Foxp3⁺ Treg were IFNγ⁺ in the PLN of liposome- than PBS-treated mice (Figure 4F, G). No differences in proportion, tetramer MFI, proportion of Foxp3⁺ or cytokine production of ChgA-specific T cells were observed in the pancreas of liposome-treated mice (Figure S3). Thus, ChgA-specific Foxp3⁻IL-10⁺ CD4⁺ T cells expanded in the spleen, suggesting regulatory potential. While ChgA-specific Tregs did not expand in the PLN, their function was modified by liposome administration.

Figure 4. Foxp3⁻ ChgA-specific CD4⁺ endogenous T cells accumulate as T_EM with regulatory-like phenotype.

Female NOD mice (8–10 weeks old) received 2 injections s.c. of 100μl liposomes encapsulating BDC2.5_mim with or without D3 at day 0 and 7. Control mice received PBS. Data compare CD4⁺ tetramer⁺ Foxp3⁻ and Foxp3⁺ ChgA-specific endogenous cells at day 11 post liposomes or PBS injections. A) Percentages of naïve (CD62L^hi CD44^low) and effector memory (CD62L^low CD44^hi) phenotypes of Foxp3⁻ and Foxp3⁺ ChgA T cells in spleen. B) Percentages of ICOS-expressing cells C) of PD1 and CD73 co-expressing Foxp3⁻ and Foxp3⁺ ChgA-specific T cells in spleen and PLN. D-G) All mice were challenged with 10μg peptide in Poly I:C/IFA emulsion at day 16 or 20 and analyzed 5 days later following PMA/ionomycin/BfA restimulation in vitro. Representative dot plots for IL-10 (D) and IFNγ (E) secretion. Relative fold increase for IL-10 (F) or IFNγ (G) production over PBS controls in spleen and PLN. n=3–10 in 3 experiments. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, one-way ANOVA.

Liposome activated ChgA-specific CD4⁺ endogenous T cells induce infectious tolerance.

To confirm the regulatory function of the cells responding to liposome treatment, NOD.SCID mice were co-transferred with splenocytes from diabetic NOD mice together with splenocytes from NOD mice pre-treated twice with BDC2.5_mim/D3 liposomes or PBS s.c. or i.v. (Figure 5A). In this setting, diabetes was significantly delayed when splenocytes from s.c. BDC2.5_mim/D3 liposome-pre-treated but not i.v. liposome or PBS-pre-treated mice were co-transferred to NOD mice. Protection required the transfer of CD4⁺ splenocytes from liposome-treated donor NOD (Figure 5B) but not IL-10 (Figure 5C, D). Expansion of antigen-specific Foxp3⁺ effector Treg s.c. but not i.v. administration of BDC2.5_mim/D3 liposomes (Figures 2 and 3) is consistent with diabetes suppression by a CD4⁺ regulatory population induced by s.c. administration of BDC2.5_mim/D3 liposomes. Furthermore, although antigen-specific CD4⁺Foxp3⁻ T cells from BDC2.5_mim/D3 liposome-treated mice do produce IL-10 in response to antigen and adjuvant, infectious tolerance in this setting does not depend on IL-10 production.

Figure 5. Liposome activated ChgA-specific CD4⁺ endogenous T cells induce infectious tolerance and diabetes protection in an accelerated model of diabetes in NOD.Scid mice.

Survival curves of A) NOD.SCID mice after splenocyte transfer from i.v. or s.c. liposome pre-treated donor NOD mice and diabetogenic splenocytes, B) NOD.SCID mice after transfer of total splenocytes or CD4⁺ or CD4⁻ sorted splenocytes from s.c. liposome pre-treated donor NOD mice and diabetogenic splenocytes. (C-D) Survival curves of NOD.SCID mice after splenocyte transfer from s.c. liposome pre-treated donor NOD mice and diabetogenic splenocytes with 4 injections of 500μg anti-IL10R blocking Ab i.p. at day 5–20 (C) or 15–28 (D). N= 5–7. *p<0.05, **p<0.01, log rank test. E-G) Female NOD mice (8–10 weeks old) received 3 s.c. injections of 100μl liposomes encapsulating BDC2.5_mim with D3 at day 0, 7 and 14. Control mice received PBS. At day 14, all mice received 10⁵ 8–3 CD90.1⁺ transgenic IGRP specific CD8⁺ T cells and PLN were then analyzed on day19 after PMA/ionomycin/BfA restimulation in vitro. Percentage (top) and count (bottom) of endogenous tetramer⁺ ChgA-specific T cells (E), transferred IGRP-specific T cells (F) and IFN-γ production by transferred IGRP-specific T cells (G). One representative experiment out of 3. (H) Percentage and count of endogenous IGRP-specific T cells on day 10 after liposomes on d0, 7. N = 4–6, *p<0.05, **p<0.01, ***p<0.001, t-test.

Since the development of T1D in NOD mice depends on islet recognition of multiple self-antigens, including IGRP-specific CD8⁺ T cells [28], we hypothesized that the liposome-induced ChgA-specific CD4⁺ regulatory population would suppress IGRP-specific effector CD8⁺ T cells in PLN, where IGRP antigen is presented by DCs draining the islets. To test this, we treated NOD mice s.c. with BDC2.5_mim/D3 liposomes or PBS on days 0, 7 and 14, transferred CD90.1⁺ IGRP-specific CD8⁺ 8.3 TCR tg T cells on day 14, then analyzed PLN 5 days later. Whereas ChgA-specific CD4⁺ T cells were expanded in BDC2.5_mim/D3 liposome-treated mice (Figure 5E), CD90.1⁺ IGRP-specific T cells were decreased in number (Figure 5F) and had a reduced capacity to produce IFNγ relative to PBS-treated mice (Figure 5G). Furthermore, two liposomes treatments at day 0 and 7 decreased the number of endogenous IGRP-specific T cells (Figure 5H). These data demonstrate that liposome immunotherapy with the BDC2.5_mim CD4⁺ ChgA mimetope promotes the regulation of naïve and antigen-experienced IGRP-specific CD8⁺ T cells in the PLN, where both ChgA and IGRP epitopes are presented by endogenous antigen-presenting cells, consistent with bystander tolerance. These data indicate that liposome treatment promotes functional regulation that suppresses disease transfer and pathogenic T cells with another islet specificity.

BDC2.5_mim/D3 liposome treatment delays diabetes progression in hyperglycemic NOD mice.

To determine whether a short-course of BDC2.5_mim liposomes could prevent diabetes development, 8-week old mice received 4, weekly s.c. (Figure 6A,B) or i.v. (Figure 6C,D) injections of PBS or liposomes encapsulating BDC2.5_mim with or without calcitriol. Four i.v. administrations of BDC2.5_mim/calcitriol liposomes significantly delayed diabetes development compared to PBS, and there was a trend towards disease acceleration with i.v. administration of BDC2.5_mim liposomes without calcitriol (Figure 6C). S.c. administration of BDC2.5_mim liposomes with or without calcitriol did not significantly delay disease onset. In the same experiments, ChgA-specific T cells were monitored in the peripheral blood of a subset of mice. Liposome treatment s.c. was associated with a temporary expansion of ChgA-specific T cells at day 5, followed by a transient increase in the proportion of Foxp3⁺ Treg at day 12 and a sustained increase in the proportion of PD1⁺ CD73⁺ ICOS⁺ Foxp3⁻ T cells until day 26 (Figure 6B). The group administered i.v. BDC2.5_mim/calcitriol liposomes was characterized by a lack of expansion of antigen-specific T cells, a prolonged increase in antigen-specific Foxp3⁺ Treg and generally low levels of PD1/CD73 co-expression by Foxp3⁻ T cells by day 26 relative to baseline levels in the blood (Figure 6D).

Figure 6. BDC2.5_mim/D3 liposome treatment in prediabetic NOD mice does not stop diabetes progression.

A,C) Survival curve of NOD mice in prevention experiment after 4 weekly s.c. (A) or i.v. (C) treatments with BDC2.5_mim/D3 liposomes (n=12, red), BDC2.5_mim liposomes (n=12, blue) or PBS (n=25, black ). *p<0.05, log rank test. B,D) Percentages of ChgA-specific CD4⁺ T cells in peripheral blood and expression of Foxp3⁺, PD1⁺/CD73⁺ and ICOS⁺ at time points post initiation of s.c. (B) or i.v. (C) treatment (n=3–4). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, one-way ANOVA. ^ indicates BDC2.5_mim/D3 and * indicates BDC2.5_mim p-values compared to PBS.

In order to test disease protection in hyperglycemic mice, we randomized mice to receive liposomes s.c. or i.v. encapsulating BDC2.5_mim/D3, BDC2.5_mim, the irrelevant peptide HEL/D3 or PBS after 2 consecutive blood glucose readings above 198 mg/dL. BDC2.5_mim (p=0.0101) s.c., BDC2.5_mim/D3 (p<0.0001) s.c., BDC2.5_mim/D3 (p=0.0008) i.v. and HEL/D3 (p=0.0034) i.v. all protected mice from diabetes progression compared to the control PBS treated group (Figure 7A and B, and Figure S4). The delay in progression after HEL/D3 liposomes was short-lived compared to the protection afforded by BDC2.5_mim containing liposomes. These data indicate that liposome treatment for 4 weeks at disease onset can delay diabetes progression for up to 6 months and that liposomes encapsulating BDC2.5_mim and calcitriol promote rapid remission.

Figure 7. BDC2.5_mim/D3 liposome treatment delays diabetes progression in hyperglycemic NOD mice.

Survival curves of NOD mice in therapeutic setting following 4 bi-weekly s.c. (A) or i.v. (B) treatments after the onset of hyperglycemia. Groups are BDC2.5_mim/D3 liposomes (n=21 and 10), BDC2.5_mim liposomes (n=9 and 9), HEL/D3 liposomes (n=3 and 10) or PBS (n=28 and 9). *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, log rank test.

Discussion

In this study, we show that the liposomal co-delivery of a single high-affinity CD4⁺ mimetope (BDC2.5_mim) of ChgA, an islet autoantigen, with the immunomodulator calcitriol induces endogenous islet-specific CD4⁺ regulatory T cells that suppress the capacity of diabetogenic splenocytes to induce diabetes and suppress effector function of bystander IGRP-specific CD8⁺ T cells. While a short course of liposomes administered between weeks 8 and 12 of age only had a small impact on diabetes incidence, treatment of hyperglycemic NOD mice for 4 weeks with s.c. or i.v. liposomes protected mice from diabetes for up to 6 months.

In NOD mice, IGRP-specific CD8⁺ T cells promote development of diabetes, provided they are high avidity, or are exposed to T cell help or inflammatory stimuli [28, 29]. Furthermore, IGRP-specific CD8⁺ T cell expansion and lytic function after 12 weeks of age predicts diabetes development [30]. Our data indicate that the function of TCR-transgenic diabetogenic IGRP-specific effector cells can be suppressed in PLN by dominant bystander tolerance, in mice treated with BDC2.5_mim/D3 liposomes. Similarly, a CD4⁺ regulatory population dominantly protected against diabetes induced by diabetogenic splenocyte transfer and in spontaneous diabetes, where the diabetogenic response is likely to be driven by multiple autoreactive T cell specificities. Given their capacity to suppress IGRP-specific effector T cells, it is noteworthy that while ChgA-specific Treg expanded in response to s.c. liposomes, treatment of NOD mice before the expected expansion of IGRP-specific CD8⁺ T cells and maturation of the lytic response (8–12 weeks of age) [30] was much less effective than treatment at onset of hyperglycemia and the peak of the lytic response. Of interest, low dose anti-CD3 mAb similarly failed to protect NOD mice from diabetes when administered between 4 and 12 weeks of age [31]. The reduced efficacy during the preclinical period and IL-10 independence of diabetes protection in transfer studies strongly suggest that the ChgA-specific regulatory cells expanded by liposomes control the presentation of antigen to islet antigen-specific CD8⁺ T cells and/or control CD4⁺ T cell help in PLN, resulting in a loss of islet-specific effector function and protection from diabetes, through bystander tolerance mechanisms [32]. Indeed the frequency of endogenous IGRP-specific CTL was significantly reduced in mice receiving BDC2.5/D3 liposomes. Possible IL-10-independent mechanisms by CD4⁺ Treg include modulation or deletion of cross-presenting DC function or competition for access to antigen, costimulation or IL-2, resulting in a loss of high-avidity effector T cells.

Autoimmune diseases, including T1D, are characterized by a lack of regulation of autoreactive T_EM cells. Previously, various strategies have been shown to restore immune regulation. For example, soluble peptide (p)-class II MHC molecules or pMHC nanoparticles were shown to induce antigen-specific CD4⁺ Treg, which suppressed T1D in an IL-10 and B cell-dependent manner, and class I pMHC nanoparticles expanded low avidity CD8⁺ memory-like regulatory T cells, which suppressed T1D in an IFN-γ-dependent manner [5, 10]. These low avidity memory-like regulatory CD8⁺ T cells were CD4-dependent. BDC2.5_mim-coupled PS/PLG nanoparticles [12] were also shown to promote the expansion of antigen-specific Treg among activated BDC2.5 TCR transgenic T cells transferred to NOD mice, which suppressed disease in a PD1 and CTLA4-dependent manner. Subcutaneously-administered liposomes encapsulating calcitriol and MHCII-restricted peptide induced antigen-specific CD4⁺ pTreg in a PDL1-dependent manner. In models of rheumatoid arthritis and vasculitis post-liposome treatment, antigen-specific CD4⁺ T cells were anergic to antigen restimulation, had low affinity for pMHC, and were depleted of differentiated effector memory T cells relative to untreated diseased mice [23]. Together these studies show that the autoimmune process driving disease can be regulated through a variety of antigen-specific strategies and regulatory mechanisms.

We co-encapsulated calcitriol as an immunomodulatory drug with the islet-derived ChgA mimetope, BDC2.5_mim. Vitamin D3 has an excellent clinical safety profile and has immunomodulatory effects [33]. Addition of vitamin D3 to DCs in vitro inhibited antigen-presentation by decreasing levels of MHC class II and CD80/86, CD40 and CD83 in an NF-kB pathway-dependent manner [34, 35]. DCs conditioned with vitamin D3 in vitro polarized CD4⁺ cells towards a Treg phenotype [36, 37]. In BALB/c and HLA-DR15 transgenic mice, co-encapsulation of peptide and calcitriol was required for induction of antigen-specific immunomodulation in vivo. In contrast, in NOD mice, s.c. administration of liposomally-encapsulated BDC2.5_mim alone was sufficient to induce antigen-specific Treg while co-encapsulation of calcitriol was necessary for disease protection in both the prophylactic and therapeutic settings after i.v. administration. The mechanisms underlying these route-specific differences are unclear, but suggest a constitutively tolerogenic environment in the skin of NOD mice, or that liposomal BDC2.5_mim particularly expands regulatory T cell clones, obviating the need for calcitriol co-encapsulation when taken up by skin dLN DCs. On the other hand, PLN DCs, including CD8⁺ cross-presenting DCs may be constitutively activated due to the ongoing autoimmune process, requiring calcitriol co-delivery with i.v. liposome administration. Indeed, i.v. administered liposomes were taken up by PLN DCs in low amounts (data not shown). Nonetheless, our data indicate that s.c. administration of liposome-encapsulated ChgA mimetope induces T cells that regulate effector T cells within the islet dLN and suppress diabetes progression, indicating that systemic peptide and calcitriol administration targeted to the pancreas or PLN is not necessary for disease control.

Treg are impaired in T1D patients and in NOD mice, both qualitatively and quantitatively [38, 39]. In contrast, high frequencies of insulin-specific Treg are associated with delays in T1D progression in children [40]. Islet-specific IL-10-producing Tr1 cells were shown to be impaired in T1D patients compared to antibody-negative first degree relatives (FDR) [41] and to be enriched in recent-onset T1D patients with better glucose and disease control [42, 43]. In our study both peripherally-induced antigen-specific Foxp3⁺ Treg and Foxp3⁻ T cells expressing high levels of effector Treg markers (ICOS, PD-1, CD73 and IL-10 in response to antigen exposure) were observed in response to s.c. liposome administration. Abrogation of the ICOS pathway in NOD neonates or BDC2.5 transgenic mice exacerbates diabetes, suggesting its role in tolerance to islet antigens [44, 45]. The PD1 marker of antigen-experience highlights the importance of active antigen presentation and response in T1D regulation. PD1-tg mice have a reduced incidence of autoimmune diabetes [46] and PD-L1 deficiency or blocking PD1 or PD-L1 in NOD mice accelerated diabetes [47, 48]. PD-1⁺ Treg cells were shown to be increased in teplizumab-treated responder patients relative to non-responders [4]. Moreover, T1D may complicate cases of cancer immunotherapy with anti-PD1 or anti-PD-L1 [49]. In human clinical trials in T1D, upregulation of Treg Foxp3 and CD25 expression were biomarkers of response to IL-2, and were associated with increased Treg responsiveness [50]. Similarly T1D patients are characterized by the accumulation of resting rather than effector Treg [51]. Thus a key goal of T1D immunotherapy is to restore functional Treg, for control of islet antigen-specific effector T cells. The current studies demonstrate that liposomes encapsulating a single high-affinity CD4⁺ mimetope of the islet autoantigen ChgA and calcitriol expand antigen-experienced pTreg with effector-memory phenotype, control IGRP-specific CD8⁺ endogenous effector T cells in pancreatic LN and control diabetes progression at the onset of hyperglycemia.

Key points.

Tolerising liposomes encapsulate a high-affinity islet CD4 mimetope and calcitriol.

Liposome administration sc at onset of hyperglycemia suppresses diabetes progression.

Mimetope-specific T cells cross-regulate IGRP-specific CD8 T cells in pancreatic LN.

Abbreviations:

ChgA

chromogranin A

DC

dendritic cell

dLN

draining lymph node

ILN

inguinal lymph node

NOD

non-obese diabetic

PLN

pancreatic draining lymph node

Tg

transgenic

T1D

type 1 diabetes

Treg

regulatory T cell