Coupling of the ZnT8 and IGRP peptides to the PSB particles
Initially, we characterised the loading of the CD8-encoded peptides onto the 500 nm PSB. We selected five HLA-A*02:01-restricted antigenic epitopes that may have a high clinical relevance in type 1 diabetes (ESM Table 2) . Peptides containing a mixture of three ZnT8 peptides (ZnT8-PSB), two IGRP peptides (IGRP-PSB) or a cocktail of these five peptides (ZnT8/IGRP-PSB) were bound to the nanoparticles using ECDI. The coupling efficiencies ranged from 65.27% to 90.48%. Between these three groups, the amount of coupled peptides was comparable and ranged from 26.11 μg/mg to 36.19 μg/mg PSB (ESM Table 3). In vitro release studies demonstrated that low levels of the peptide are released at 2 h for IGRP, ZnT8 and ZnT8/IGRP-PSB (12.21%, 23.37% and 15.05%, respectively). However, a greater amount of IGRP-PSB was released at 72 h in vitro compared with that coupled to ZnT8-PSB or ZnT8/IGRP-PSB (Fig. 1a).
Individual peptide-PSB had similar coupling efficiencies to that of the mixture peptides-PSB, ranging from 52.83% to 93.19% with comparable amounts of peptide (21.13–37.28 μg/mg PSB) (ESM Table 4). Approximately half of the loaded peptide was released from ZnT8 or IGRP peptide-coupled nanoparticles by 48 h in vitro, with higher amounts of the peptide released by 72 h (Fig. 1b).
Peptides-PSB ablate antigen-specific CD8+ T cell responses in HHD II mice
We have previously demonstrated that the novel epitopes of ZnT8-3, ZnT8-6 and ZnT8-9 could generate antigen-specific CD8+ T cells in HHD II mice, and all of these epitopes are dominant in humans . Thus, we measured the capability of 500 nm PSB coupled with these immunodominant epitopes (ZnT8-PSB) to prevent ZnT8-specific T cell responses in vitro in HHD II mice. Antigen-specific CD8+ T cells were significantly increased in all groups with the related ZnT8-peptide stimulation (Fig. 2a–c). Compared with control mice, CD8+ T cells from HHD II mice incubated with ZnT8-PSB displayed significantly decreased responses to the re-stimulation induced by the ZnT8-3, ZnT8-6 and ZnT8-9 epitopes in vitro (Fig. 2a–c). Representative IFN-γ ELISpot results are shown in ESM Fig. 1. A previous study has indicated that Ag-SPs are differentially responsible for the diabetes-protective effects . Hence, we chose to couple ZnT8 and two other immunodominant antigen epitopes (IGRP-1 and IGRP-2), which were demonstrated in our human participants, with PSB to determine whether the peptides-PSB complexes were responsible for the different protective immunity effects in vivo. Because peptides-PSB could induce long-term T cell tolerance , ZnT8-PSB and IGRP-PSB (mixture of IGRP-1 and IGRP-2) were administered intravenously to HHD II mice immunised with an adjuvant mixture of the five ZnT8 and IGRP epitopes. Compared with control mice that received cytomegalovirus (CMV)-PSB, the frequencies of both CD4+ and CD8+ T cells were slightly reduced in the mice treated with ZnT8-PSB and IGRP-PSB. In addition, the number of T helper (Th) 1 cells was decreased in the HHD II mice treated with PSB. However, no significant differences were found in the number of Th17 cells in these treated mice (Fig. 2d). Treatment with ZnT8-PSB and IGRP-PSB also significantly decreased the activated CD4+ T cell (CD62LlowCD44high) responses. Although the activated CD8+ T cell (CD62LlowCD44high) decreased in the mice treated with peptides-PSB, no statistically significant differences were found among the groups. Conversely, the concentration of naive CD4+ and CD8+ T cells (CD62LhighCD44low) was significantly higher in the mice treated with peptides-PSB than in their control mice (Fig. 2e).
In the HHD II mice treated with ZnT8-PSB or IGRP-PSB, antigen-specific CD8+ T cell responses to both the HLA-A*02:01-restricted ZnT8 and IGRP peptides were significantly decreased compared with control mice (Fig. 2f–j). Interestingly, treatment with ZnT8-PSB showed a more significant reduction in CD8+ T cell responses to the IGRP-1 peptide than treatment with IGRP-PSB (Fig. 2f). These in vivo assays further demonstrated that the HHD II mice treated with peptides-PSB are tolerant to the peptides, as shown by reduced T cell frequencies and IFN-γ secretion following a challenge with the ZnT8 and IGRP peptides.
Peptides-PSB induce Tregs and IL-10 production and abort T cell proliferation
A previous study showed that the induced T cell tolerance was partially dependent on the production of antigen-specific Tregs and IL-10 . The FoxP3+ Tregs were significantly increased after treatment with either ZnT8-PSB or IGRP-PSB in immunised HHD II mice (Fig. 3a). Treatment with ZnT8-PSB and IGRP-PSB resulted in a significantly higher increases in IL-10 cytokines in response to the IGRP or ZnT8 peptides (Fig. 3b–f). Representative IL-10 ELISpot results are shown in ESM Fig. 2. Altogether, these data indicate that both ZnT8-PSB and IGRP-PSB are significantly important for inducing antigen-specific CD8 T cell tolerance in HHD II mice.
To further investigate the effects of the peptides-PSB on T cell activation and proliferation, we immunised HHD II mice with the ZnT8-3, ZnT8-6 or ZnT8-9 epitope. Then, we adoptively transferred carboxyfluorescein diacetate succinimidyl ester (CFSE)-labelled CD8+ T cells from these immunised mice into naive HHD II mice. After 48 h, we administered 9 × 109 ZnT8-PSB or CMV-PSB via i.v. injection. After 5 days, we administered ZnT8-3, ZnT8-6 or ZnT8-9 plus complete Freund’s adjuvant (CFA) via s.c. injection. The ZnT8-specific T cells isolated from both the spleen and lymph nodes of the mice treated with ZnT8-PSB exhibited a significantly lower proliferative capacity than those from the control mice (Fig. 3g).
Peptides-PSB induces CCL22 expression specifically in CD169+ MMΦ
Splenic function plays an important role in systemic tolerance by clearing apoptotic cells . Splenic macrophages (MΦs) are involved in the rapid phagocytosis of different pathogens and other particles . Previous studies have also demonstrated that treatment with peptides-PSB increases the expression of some scavenger receptors . We found that the scavenger receptors Marco and Scarf-1 were upregulated in the spleen following the administration of ZnT8-PSB (Fig. 4a, b). However, no significant difference was found in the expression Sign-r1 (Fig. 4c). The spleen cells containing ZnT8-PSB also induced the expression of Siglec-1, which is known as a metallophilic macrophage (MMΦ) marker (Fig. 4d). Overall, the data indicate that i.v. administration of peptides-PSB specifically localises to macrophage receptor with collagenous structure (MARCO)+ marginal zone macrophages (MZMΦs) and MMΦs. These MΦs act as scavenger cells, developing anti-inflammatory responses.
In the spleen, MZMΦs and marginal zone (MZ) DCs actively phagocytose apoptotic cells from circulation . After activation, both MΦs and DCs can produce CCL22 . CCL22 also plays a key role in Treg recruitment and immune suppression . When FACS-sorted splenic phagocytes (see ESM Fig. 3) were examined, only CD169+ MMΦs and CD8+ DCs showed an increase in the expression of Ccl22 (Fig. 4e). In CD169+ MMΦs, the CCL22 response is specific and unique to apoptotic cell exposure . Our data also confirm that peptides-PSB did not induce significant expression of C-C motif chemokines other than Ccl22 in the CD169+ MMΦs (Fig. 4f).
Peptides-PSB induce rapid DC and Treg accumulation in the MZ
Apoptotic cell-driven MZMΦ activation potently stimulated Treg migration. This response was dependent on CCL22-mediated recruitment via CCR4 . To determine whether this mechanism is relevant to the administration of ZnT8-PSB, we challenged mice with an injection of a CCR4 antagonist after immunisation with a ZnT8 cocktail and then examined FoxP3+ Treg accumulation 4 h after pretreatment with ZnT8-PSB. Mice injected with ZnT8-PSB exhibited significantly more splenic FoxP3+ Tregs than the control mice, and this effect was completely inhibited by a treatment with the CCR4 antagonist (Fig. 5a). Treg recruitment primarily occurred in the splenic MZ of the mice treated with ZnT8-PSB (Fig. 5b), suggesting that CCL22 might increase Treg chemotaxis to the spleen, promoting the increased expression of Tregs in the organ. Pretreatment with the CCR4 antagonist prevented this increase in Tregs, resulting in splenic FoxP3 staining that was similar to that in the control groups (Fig. 5b, c). Meanwhile, the CCL22- and CCR4-dependent follicular accumulation of CD11c+ cells was recruited from the MZ rather than systemically (Fig. 5b, d). Moreover, image analysis showed that there was a significant increase in follicular Treg–DC interactions in the follicle in the mice treated with ZnT8-PSB (Fig. 5b, e). In response to peptides-PSB, splenic Tregs rapidly upregulated surface cytotoxic lymphocyte (CTL) A-4, whereas CD103 and programmed cell death-1 (PD-1) expression was not changed (Fig. 5f). Moreover, pretreatment with the antagonist prevented surface CTLA-4 expression (Fig. 5f). Furthermore, when the MMΦs were cultured with apoptotic CD8+ T cells treated with ZnT8-PSB, the conditioned media induced Treg migration and this effect was inhibited by adding either a CCL22-neutralising antibody or the CCR4 antagonist (Fig. 5g). CD8+CD103+ DCs preferentially present apoptotic cell-associated antigen in the spleen and induce Treg differentiation and promote T cell tolerance [34, 35]. We found that CCR4 was significantly more highly expressed in the CD103+ DCs than in the CD103− DCs in the mice treated with ZnT8-PSB (Fig. 5h), suggesting that CD8+CD103+ DCs may preferentially migrate in response to CCL22 production. Consistent with this result, we found that interactions between apoptotic CD8+ T cells and MMΦs mediated a significant migration of CD8+CD103+ DCs in a CCL22- and CCR4-dependent manner (Fig. 5i). In contrast, CD8+CD103− DCs were independent of CCL22/CCR4 and did not show an apoptotic cell-dependent migratory capacity (Fig. 5i).
Because CCL22 expression increased rapidly after the peptides-PSB administration and was a driver of Treg accumulation, we hypothesised that regulatory cytokine production may be based on CCL22-mediated Treg recruitment. We found that ZnT8-PSB prominently increased splenic TGF-β and IL-10. When CCR4 was blocked, the induction of IFN-γ and IL-12 was significantly increased (Fig. 6a, b), and the induction of TGF-β and IL-10 was significantly decreased (Fig. 6c, d). These findings suggest that deficiency in the CCL22-dependent Treg recruitment led to the formation of a proinflammatory environment.
We further confirmed whether CCL22-mediated Treg recruitment could promote the prevention of inflammatory cytokine production and the acquisition of a regulatory phenotype in MΦs and DCs following peptides-PSB treatment. Tgf-β transcription increased in CD169+ MΦs (p = 0.0038) and CD8+ DCs (p = 0.0194) (Fig. 6g, k) after a ZnT8-PSB challenge. Similarly, the Il-10 expression was increased in the CD169+ MΦs (p = 0.0006) and CD8+ DCs (p = 0.0017) following i.v. injection of ZnT8-PSB (Fig. 6h, l). However, when CCR4 function was inhibited by the pretreatment with an antagonist, the peptide-PSB-driven increases in Tgf-β and Il-10 mRNA levels were blocked in both CD169+ MΦs (p = 0.0015, p = 0.0437, respectively) and CD8+ DCs (p = 0.0362, p = 0.0020, respectively) (Fig. 6g, h, k, l) with a concomitant significant increase in Il-12p40 and Il-6 in CD169+ MΦs (p < 0.0001, p = 0.0059, respectively) (Fig. 6e, f, i, j).
Peptides-PSB prevent autoimmune diabetes in NOD.β2m
null HHD mice
To further assess whether peptides-PSB would interfere with disease development, we administered ZnT8-PSB, IGRP-PSB and ZnT8/IGRP-PSB to 8-week-old NOD.β2m
null HHD mouse models of diabetes. In the control groups, the mice developed diabetes at 12 weeks of age. The incidence of diabetes gradually increased over the treatment period. In contrast, only 12.5~18.75% of mice receiving peptides-PSB developed diabetes during the treatment period (Fig. 7a). Moreover, treatment with ZnT8-PSB and ZnT8/IGRP-PSB significantly delayed the development of diabetes and improved survival rates (Fig. 7a, b). Although the mice treated with IGRP-PSB similarly reduced the incidence of diabetes, the difference was not statistically significant (Fig. 7a). Treatment with IGRP-PSB only improved survival rates (Fig. 7b). The protection observed in the animals treated with peptides-PSB was accompanied by diminished islet infiltration (Fig. 7c, d) as determined by histological examination. In contrast to control mice, the number of FoxP3+ Treg cells was significantly increased in the pancreas after treatment with peptides-PSB (Fig. 7e). The mice receiving peptides-PSB showed a reduced degree of CD4+ and CD8+ T cell infiltration (Fig. 7f). It must be noted here that only non-diabetic mice in these groups were analysed at 28 weeks of age. These results show that treatment with peptides-PSB is essential for ameliorating insulitis and delaying disease progression in mouse models of type 1 diabetes.
Peptides-PSB ablate antigen-specific CD8+ T cell responses in individuals with type 1 diabetes
The peptides-PSBs were subsequently tested for antigen-specific T cell responses using peripheral blood mononuclear cells (PBMCs) from individuals with HLA-A*02:01-restricted new-onset type 1 diabetes who were previously positive for these epitopes. PBMCs were co-cultured in vitro with ZnT8-PSB, IGRP-PSB or ZnT8/IGRP-PSB, and antigen-specific CD8+ T cells for each individual epitope were detected by ELISpot assays. Compared with those exposed to the control human immunodeficiency virus (HIV)-PSB, CD8+ T cells co-cultured with ZnT8-PSB displayed significantly diminished responses to the re-stimulation with the ZnT8-3, ZnT8-6 and ZnT8-9 epitopes (Fig. 8a–c). Similarly, IGRP-PSB significantly decreased the ability of CD8+ T cells to respond to re-stimulation with IGRP-1 and IGRP-2 (Fig. 8d, e). Furthermore, the co-incubation with ZnT8/IGRP-PSB diminished the CD8 T cell responsiveness to the IGRP, ZnT8-3 and ZnT8-9 epitopes (Fig. 8f). Although the ZnT8-6 epitope-specific T cell responses were reduced by ZnT8/IGRP-PSB treatment, the effect did not quite achieve statistical significance (Fig. 8f).