Soluble Antigen Arrays Efficiently Deliver Peptides and Arrest Spontaneous Autoimmune Diabetes Free

Autoimmune diseases, including type 1 diabetes (T1D), are manifestations of tissue damage caused by autoreactive lymphocytes that have eluded mechanisms of central and/or peripheral tolerance. Antigen-specific immunotherapy (ASIT) aims to expose immune cells to relevant antigens under noninflammatory conditions to tolerize/desensitize specific populations of autoreactive lymphocytes (1). Depending on conditions of exposure (e.g., route, dose, and antigen-presenting cells [APCs] involved), engagement of autoreactive lymphocytes after ASIT can achieve antigen-specific tolerance through deletion, induction of anergy, and/or conversion to regulatory/suppressive populations.

Peptide-based ASITs focus on relevant epitopes, allowing exclusion of unnecessary or unwanted epitopes and inclusion of epitopes from multiple antigens. However, peptides tend to disperse rapidly, requiring relatively high doses to achieve proper engagement of T cells in vivo. Another limiting factor is that several peptides have been shown to cause anaphylactic reactions after repeated administration (2–5), and NOD mice seem more susceptible than other common strains (4). Issues of dispersion, requirement for high doses, and risk of anaphylaxis could be mitigated by using proper delivery systems that facilitate drainage to lymphoid tissues, control release of peptides, and minimize their degradation (6,7). Soluble antigen arrays (SAgAs) constitute a new class of ASIT comprising multiple copies of peptides chemically conjugated onto small hyaluronic acid (HA) chains and providing soluble, multivalent, and linear presentation and delivery of these peptides. Increased molecular weight minimizes wide dispersion into systemic circulation, which can occur for peptides, and instead facilitates drainage to lymph nodes (LNs). Their solubility and structure also distinguish SAgAs from particulate forms of delivery vehicles that characterize conventional (immunogenic) vaccines. The multivalent display of antigens by SAgAs enables high-avidity interactions (e.g., proteins linked on SAgAs with B cells), resulting in more effective anergy induction (8). Two forms of SAgAs were evaluated in this study, one in which the linkers for grafted peptides are hydrolysable (hSAgAs) and one in which the peptides are more stably linked by click chemistry (cSAgAs) (9).

We previously showed that SAgAs can ameliorate disease in the experimental autoimmune encephalomyelitis model of multiple sclerosis (10) and that SAgAs can efficiently engage diabetogenic T cells (9). In both patients with T1D and NOD mice, the list of targeted epitopes has been greatly extended from native epitopes to neoepitopes, including posttranslationally modified versions of native epitopes and hybrid peptides (11,12). These recently identified neoepitopes likely play a critical role in the pathogenesis of this disease (13,14). Indeed, these neoepitopes may not be presented during thymic selection of autoreactive T cells, and reactive T cells may therefore circulate at higher frequencies. Moreover, many diabetogenic T cells have a low affinity for native epitopes and may be more strongly engaged by modified epitopes (15) or alternatively by mimotopes (16,17). The therapeutic utility of hybrid insulin peptides (HIPs) and mimotopes has been demonstrated (16–18). In this study, we provide evidence that two SAgAs displaying 2.5HIP (11) and the p79 mimotope (19) are able, in combination, to prevent the onset of disease in NOD mice and that SAgAs, particularly cSAgA, make the delivery of peptides more effective and potentially safer.

We purchased HA sodium salt (molecular weight 16 kDa) from Lifecore Biomedical; 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride and N-hydroxysuccinimide from TCI Chemicals; 2-(N-morpholino)ethane-sulfonic acid sodium salt, tris(3-hydroxypropyltriazolylmethyl)amine, and sodium ascorbate from Sigma-Aldrich; and 11-azido-3,6,9-trioxaundecan-1-amine, N-hydroxysuccinimide, and copper (II) sulfate pentahydrate from Acros Organics. Alkyne-functionalized peptides bearing an N-terminal 4-pentynoic acid (homopropargyl) modification, hpP79 and aoP79, were obtained from Biomatik. Additionally, alkyne-functionalized peptides bearing an N-terminal alkyne polyethylene glycol group modification, 2.5HIP and a control p79 with two permuted amino acids inactive on BDC2.5 T cells, were purchased from PolyPeptide. All reagents were used as received without further purification.

SAgAs were synthesized by conjugating ∼10 peptides to a 16-kDa HA backbone. SAgAs used either a hydrolysable (fSAgA, hSAgA_p79, and hSAgA_2.5HIP) or nonhydrolysable (cSAgA_p79 and cSAgA_2.5HIP) linker and were produced as previously described (10,20). For uptake studies, FITC was conjugated directly to HA using stable click linker chemistry (two FITC molecules on average per SAgA molecule for both hSAgA_p79 and hSAgA_2.5HIP). Azide-functionalized HA, hSAgAs, and cSAgAs were characterized by nuclear magnetic resonance, dynamic light scattering, and circular dichroism, as previously reported (9). Peptide conjugation was determined through gradient reversed-phase high-performance liquid chromatography. Initial studies were focused on cellular effects using the hSAgA format; comparative studies of hSAgAs and cSAgAs then revealed a surprising enhancement of potency with cSAgAs, with greater disease protection in later preclinical studies. This motivated our studies exploring mechanistic differences between peptides in free form, those releasable from HA via hydrolysis (hSAgAs) and those more durably linked to HA (cSAgAs).

NOD (#001976; Jax), BDC2.5 (#004460; Jax), NOD.CD45.2 (#014149; Jax), and NOD.Foxp3/GFP mice (#025097; Jax) were obtained from The Jackson Laboratory. NOD.Foxp3/GFP mice were bred with BDC2.5 mice and NOD.CD45.2 mice to produce BDC2.5.Foxp3/GFP mice with CD45.2 and/or CD45.1 congenic markers for in vivo tracking. All mice were bred and maintained in the specific pathogen-free animal barrier facility of the Columbia Center for Translational Immunology. Both male and female transgenic/congenic mice (6–16 weeks of age) were used as donors for adoptive transfer experiments. Male and female NOD mice (8–12 weeks of age) were used as recipient mice for adoptive transfer and in vivo uptake studies. Female NOD mice (8 weeks of age) were used in studies involving treatments (preclinical) and MHC tetramers (Tetrs). All procedures were approved by the Columbia University Institutional Animal Care and Use Committee and performed in adherence with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Mice were treated with subcutaneous injection of FITC-conjugated hSAgA_p79 and hSAgA_2.5HIP in combination or separately. Several LNs and spleen were analyzed 6 and 12 h later for FITC signal in various immune populations. For in vitro uptake studies, splenocytes were analyzed at indicated time points after culture with FITC-conjugated hSAgAs. Cell populations were analyzed on a BD Fortessa using a combination of markers for major APC populations (Supplementary Table 1).

In all studies, SAgA_mix indicates a mix of two different SAgAs, each with one type of peptide. In preclinical studies, mice were treated with saline, HA, free peptide with or without HA, hSAgA, or cSAgA by subcutaneous injection (behind the neck) weekly starting at 8 weeks of age, and blood glucose levels were monitored weekly. Mice were considered diabetic when their glycemia exceeded 250 mg/dL in two consecutive readings 2 days a part. In adoptive transfer studies, mice were treated subcutaneously with a single dose of saline, free peptide(s), hSAgA, or cSAgA. Free peptides and SAgAs were always injected based on the same equimolar amount of peptides (1 nmol SAgA ∼10 nmol peptides). Donor CD4⁺ CD25⁻ T cells from pooled LNs and spleen of BDC2.5 or BDC2.5.Foxp3/GFP mice (with CD45.2 congenic marker) were purified using the MojoSort Mouse CD4 T Cell Isolation Kit supplemented with biotinylated anti-CD25 (BioLegend) and then labeled with Violet Cell Proliferation Dye (eBioscience) as previously described (9). These antigen-specific T cells (0.5 × 10⁶ to 1 × 10⁶ cells) were injected intravenously on the same day as the treatment in all adoptive transfer studies, except in studies assessing persistence of antigen presentation, in which T cells were injected 1, 2, or 3 days after treatment with peptide or SAgA. Three days after T-cell transfer, spleen and various LNs were collected for analysis of T-cell responses by flow cytometry to assess surface markers and intracellular cytokines (Supplementary Table 2), along with GFP (Foxp3) and proliferation. Intracellular staining was performed after a 4-h incubation with PMA (0.1 µg/mL), ionomycin (40 µg/mL), brefeldin A (1.5 µg/mL), and monensin (1 µmol/L). To assess endogenous antigen-specific T-cell responses, NOD mice were treated with SAgAs or saline, and responding T cells were identified with I-A^g7/p79-Tetrs and/or I-A^g7/2.5HIP-Tetrs (Supplementary Table 3) after staining for 1 h at 37°C.

Statistical analysis was performed using GraphPad Prism’s two-way ANOVA with Tukey post hoc tests or unpaired t tests, as indicated in each figure legend. Log-rank tests were used in studies of diabetes and anaphylaxis incidence. The threshold for statistical significance was P < 0.05 (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Flow cytometric data were analyzed with FCS Express 7.

The data sets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request. Research resource identifiers for antibodies are provided in Supplementary Tables 1 and 2.

To assess uptake of SAgAs in vivo, we injected a mix of FITC-conjugated hSAgA_p79 and hSAgA_2.5HIP and assessed relative uptake by different immune cells in draining and distal LNs and spleen. SAgAs were clearly detected in subsets of migratory and resident CD11c⁺ dendritic cells (CD11b⁺ and CD8a⁺) as well as in other myeloid populations (CD11b⁺ CD11c⁻) within 6 h (Fig. 1A and Supplementary Fig. 1A and B), but to a limited extent in other populations (data not shown). The sequence of the two peptides predicted lower solubility of SAgA_2.5HIP. To determine how this may influence uptake and in vivo diffusion, we compared the relative uptake of the two different SAgAs in vitro and in vivo. Uptake of SAgA_p79 was consistently greater than that of SAgA_2.5HIP at two time points and in the above three APC populations in vitro (Supplementary Fig. 1C and D). The same consistent difference was observed in draining LNs in vivo (Fig. 1B). In all in vivo studies, uptake was highest in draining axillary and brachial LNs followed by cervical LNs (draining more remotely) and lowest in mesenteric and pancreatic LNs (distal) and spleen, indicating a gradient of biodistribution to lymphoid tissues.

Several T1D-relevant antigenic peptides were considered for SAgA formulation based on known reactivity of diabetogenic T cells, but some of them (e.g., InsB:9–23) were not advanced in SAgA form because of insufficient solubility. p79 and 2.5HIP were among the top candidate peptides that passed our selection criteria and were validated (9) for further investigation. Two forms of SAgAs (hSAgAs and cSAgAs, with peptides grafted using hydrolysable and nonhydrolysable linkers, respectively) were compared against equimolar amounts of free peptides for T-cell responses elicited in the aforementioned lymphoid tissues. NOD mice received BDC2.5 CD4⁺ T cells, which react to both p79 and 2.5HIP peptides, and were treated with these two peptides in free form or one of the SAgA forms. Both forms of SAgAs induced greater clonal expansion, which was highest in draining LNs followed by distal LNs and spleen (Fig. 2A and B; only cSAgAs at lower doses), and a higher percentage of proliferated cells (Fig. 2C and D) compared with free peptides. At a 0.1-nmol dose, cSAgAs induced more proliferation and CD44 upregulation than hSAgAs; this difference disappeared at a 0.5-nmol dose (Fig. 2A–F). Surprisingly, CD44 was either less induced or downregulated in response to a 0.5-nmol dose of SAgAs relative to peptides (Fig. 2F). CD25 induction was most prominent with SAgAs, but only in draining LNs and at a 0.5-nmol dose (Supplementary Fig. 2A and B).

Both SAgAs induced checkpoint molecules Lag-3 (Supplementary Fig. 2C and D) and programmed death-1 (PD-1) (Supplementary Fig. 2E), whereas peptides induced none or few of these molecules (Supplementary Fig. 2F–H). Moreover, SAgAs induced pronounced upregulation of CD73, both in terms of percentage and level (Fig. 2G and H and Supplementary Fig. 2G), whereas free peptides did not. CD73 induction was consistently higher with cSAgAs than with hSAgAs. The difference in percentage of CD73⁺ FR4⁺ cells (defined as anergic) (21,22) was even more dramatic between free peptides and SAgAs (Supplementary Fig. 2I). Treatments had no major effects on recipient T cells (e.g., as shown for CD73/FR4) (Supplementary Fig. 2J), although for some measurements, we did observe some marginal upregulations resulting from the presence of responding antigen-specific T cells among recipient cells. Thus, SAgAs induce greater expression of tolerance-related markers CD73/FR4, Lag-3, and PD-1 than free peptides in vivo, in a manner consistent with the gradient of antigen uptake.

We assessed the potential of SAgAs to induce regulatory T-cell (Treg) populations, which consist primarily of Foxp3⁺ and interleukin-10⁺ (IL-10⁺) Tregs (also known as Tr1). The BDC2.5 CD4⁺ T cells used in most adoptive transfer experiments carried the Foxp3/GFP reporter. Both free peptides and SAgAs induced Foxp3 expression in donor T cells in a dose-dependent manner (Fig. 3A and B and Supplementary Fig. 3A). Foxp3 was upregulated only at the lower dose (0.1 nmol), and this was most pronounced at distal sites (Fig. 3A), where the dose was even lower, according to the antigen gradient. Similar effects were seen with treatment using 2.5HIP peptide or hSAgA_2.5HIP only (Supplementary Fig. 3B and C); proliferation was comparable in all tested tissues (Supplementary Fig. 3D). We also assessed interferon-γ (IFN-γ) and IL-10 expression by responding T cells, which was detected only at the higher dose (0.5 nmol). SAgAs increased expression of IL-10 but not of IFN-γ relative to free peptides in donor T cells (Fig. 3C and Supplementary Fig. 3E), as well as in recipient T cells marginally in the spleen (Fig. 3D). Most IL-10⁺ T cells did not coexpress IFN-γ (Fig. 3C). As a result, the ratio of IL-10 to IFN-γ was significantly higher with SAgAs than with free peptides in antigen-specific CD4⁺ T cells (Fig. 3E) but was not changed in recipient polyclonal CD4⁺ T cells (Fig. 3F). In sum, SAgAs induced Foxp3⁺ Tregs at a lower dose and in distant lymphoid tissues and IL-10⁺ Tr1 cells at a higher dose, in which case SAgAs were significantly more efficient than the same amount of peptides in free form.

We wondered if the differences in T-cell responses between SAgAs and free peptides could be linked to more prolonged presentation of peptides, indirectly reflecting an extended half-life of the peptides in vivo. Given the superior therapeutic potential of cSAgA (presented next), we compared free peptide versus cSAgA administration. To detect antigen presentation, antigen-specific donor T cells were transferred 1, 2, or 3 days after treatment with free peptide or SAgA at equivalent dose. Antigen presentation from free peptides decreased significantly each day based on proliferation and CD44 upregulation of donor T cells, whereas antigen presentation from SAgAs remained stable for at least 3 days in draining LNs (Fig. 3G and H). As a control, there was no difference and no change over time in pancreatic lymph nodes, where the presented antigen was predominantly from islets (Fig. 3I).

Our preclinical evaluation of SAgAs to prevent disease in NOD mice was first done with hSAgAs. Treatment with a mixture of hSAgAs (hSAgA_mix; p79 and 2.5HIP at 2.5 nmol each) demonstrated complete protection until 18 weeks of age, whereas more than half of the control mice had diabetes by that time (Fig. 4A). However, some hSAgA_mix-treated mice started to develop anaphylaxis at 18 weeks (fatal for three mice), and the treatment was discontinued. tAfter treatment discontinuation, protection was gradually lost, and mice developed diabetes, although with a significant delay relative to controls. Here, HA alone had a marginal protective effect, which was not seen in later experiments (e.g., Fig. 4B). In a follow-up study, we added a group with a lower dose (0.5 nmol of each hSAgA) and longer treatment period. Mice treated with 0.5 nmol hSAgA_mix never developed anaphylaxis and had partial but not significant protection from disease, with half of the mice remaining diabetes free at 33 weeks (Fig. 4B). Incidence of disease remained low (70% mice were diabetes free by 36 weeks of age) in the 2.5-nmol group compared with saline and HA controls, suggesting that dose and prolonged dosing were important for long-lasting protection. However, anaphylaxis caused the loss of two mice at 16 weeks, and dosing was halved (Fig. 4B). At 22 weeks, anaphylaxis caused the loss of another three mice, at which point the treatment was ceased. In contrast, each single hSAgA tested separately (2.5 nmol each) had no significant effect on disease incidence on their own and did not induce anaphylaxis (Fig. 4C). Reminiscent of previous reports on other peptides, our peptide mix elicited an anaphylactic reaction, and we therefore ascertained whether the SAgA form of delivery was responsible. We treated NOD mice with a mix of free peptides, hSAgAs, or cSAgAs (all with p79 and 2.5HIP peptides), with dose tapering from 2.5 nmol of each SAgA to 1.5 nmol after five injections and to 0.5 nmol after four injections. Incidence of anaphylaxis was more severe with free peptides than with hSAgA_mix (Fig. 4D), whereas cSAgA_mix did not induce any lethal anaphylaxis (Fig. 4D). Mice treated with cSAgA_mix did not develop diabetes during dosing at 2.5–1.5 nmol and had significantly reduced incidence of hyperglycemia during 0.5-nmol dosing, with 70% of mice diabetes free and healthy at 30 weeks of age (Fig. 4E). Thus, not only was SAgA not responsible for peptide-induced anaphylaxis, it minimized anaphylaxis. Overall, cSAgA treatment was most effective clinically while averting these adverse effects.

To address the immunologic effects of SAgAs on endogenous (polyclonal) antigen-specific T cells in NOD mice, we used MHC Tetrs to analyze p79-reactive CD4⁺ T cells first. Their frequency was significantly increased after multiple doses, ranging from two to six doses (Supplementary Fig. 4A and B showing results after six doses of hSAgA_mix), and they persisted as late as 21 weeks after treatment was discontinued (Supplementary Fig. 4C and D). These antigen-specific T cells displayed marked upregulation of CD73 and FR4 in response to six weekly doses of hSAgA_mix, particularly with the 2.5-nmol dose relative to 0.5 nmol (Fig. 5A and B). Similarly, expression of PD-1 was also significantly increased (50% in treated mice as compared with 5% in control mice) (Fig. 5C). These effects were also evident after three weekly doses of SAgA_p79 (Fig. 5D–F); however, cSAgA_p79 was more potent than hSAgA_p79 at inducing CD73/FR4 (Fig. 5D and E) and PD-1 (Fig. 5F). CD73 upregulation persisted after 21 weeks of treatment interruption (Supplementary Fig. 4E). At the earliest time point tested (after two injections, 3 days apart), the effect of hSAgA_p79 on upregulation CD73/FR4 on p79-reactive T cells was limited, whereas that of cSAgA_p79 was already pronounced (Supplementary Fig. 4F and G). Although nonspecific T cells preferentially express IFN-γ over IL-10, p79-reactive T cells comprised more IL-10⁺ than IFN-γ⁺ cells after six weekly treatments (Fig. 6A–C). Consistent with surface markers, frequency of cells expressing these cytokines increased with dose of hSAgAmix (Fig. 6B). After only three weekly treatments, the effect of hSAgA_p79 relative to control mice was not yet significant (Fig. 6D–F), whereas cSAgA_p79 had induced significantly more IL-10 single-positive cells among p79-reactive T cells than hSAgA_p79 in both LNs and spleen (Fig. 6E). The increase in IL-10 single-positive p79-Tetr⁺ cells was most pronounced in LNs.

The p79 and 2.5HIP peptides share similar COOH-terminal residues (LWVRME for p79 and LWSRMD for 2.5HIP), which are recognized by the BDC2.5 T-cell clone. Using an in vitro stimulation assay with NOD splenocytes spiked with BDC2.5 splenocytes (9), we found that hSAgA_2.5HIP was ∼10-fold less active than hSAgA_p79 (Supplementary Fig. 5). We addressed whether endogenous p79- and 2.5HIP-reactive T cells were generally cross-reactive like the BDC2.5 clone. Using both p79-Tetr and 2.5HIP-Tetr, we tracked the response of both populations in response to cSAgA_p79, cSAgA_2.5HIP, and cSAgA_mix. First, p79-Tetr⁺ and 2.5HIP-Tetr⁺ T cells were largely nonoverlapping, and few T cells costained for both Tetrs, regardless of treatment (Fig. 7A). Second, clonal expansion was only significant for p79-Tetr⁺ T cells in response to p79 alone or in the mix, but not to 2.5HIP alone (Fig. 7B), suggesting lack of cross-reactivity. Third, consistent with previous experiments, cSAgA_p79 strongly induced CD73⁺ cells (Fig. 7C and D) and CD73⁺ FR4⁺ cells (Supplementary Fig. 6A and B). Upregulation of CD73/CD44 (Fig. 7C and D) and KLRG1 (Fig. 7E and F) was evident on both p79- and 2.5HIP-reactive T cells in response to their respective epitope, confirming that both populations were effectively and specifically stimulated by their respective antigens, although cSAgA_p79 was more potent. Upregulation of PD-1 was also evident in p79-Tetr⁺ T cells, without reaching significance (Supplementary Fig. 6C). The response of p79-reactive T cells was dominated by IL-10⁺ cells, whereas that of 2.5HIP-reactive T cells did not change and, in the spleen, was largely dominated by IFN-γ⁺ cells (Fig. 8). In 2.5HIP-specific T cells, a majority of KLRG1⁺ cells expressed IFN-γ, whereas in p79-reactive T cells, they did not (Supplementary Fig. 6D). No significant de novo induction of Foxp3⁺ cells among p79- and 2.5HIP-reactive T cells was seen; however, stimulation of existing Foxp3⁺ T cells (based on CD44 coexpression) was significant with cSAgA_mix, but not with the free peptide mix (Supplementary Fig. 7A and B). The free peptide mix had no significant effect on IL-10 induction or ratio of IL-10 to IFN-γ, unlike cSAgA_mix (Supplementary Fig. 7C and D). Finally, an effect on the frequency and stimulation of CD4⁺ T cells specific to other β-cell antigens was not seen (Supplementary Fig. 8).

SAgAs constitute a novel and effective antigen delivery platform that is soluble and avoids the pitfalls of particle-based approaches. The solubility and size of SAgAs promote drainage to lymphoid tissue, and their multivalency disarms autoimmune cells, making SAgAs uniquely suitable for tolerance induction (6,8,10,23,24). In fact, structural features of SAgAs harken back to seminal work pointing to soluble multimeric haptens <100 kDa as inhibitors of immunity and larger or insoluble multimers as instigators (25). Here, we show that SAgAs promote more effective and persistent engagement of autoreactive T cells than free peptides in vivo, particularly cSAgAs. This is in part attributed to the higher molecular weight of SAgAs, which enables more efficient drainage to lymphoid tissues. T-cell responses correlated with the extent of SAgA uptake, highest in draining LNs and lower in more distal lymphoid tissues. One exception was the induction of Foxp3, which was greatest with lower administered doses and at more distal sites. This observation is consistent with lower (subimmunogenic) doses of insulin mimotopes that preferentially induced Foxp3⁺ Tregs (16,17). Over a certain threshold of stimulation, Foxp3 is no longer induced, and IL-10⁺ Tr1 cells seem to take over. However, at a dose supporting IL-10 induction, cSAgAs (but not free peptides) can stimulate existing p79- and 2.5HIP-reactive Foxp3⁺ CD4⁺ T cells, which may contribute to overall protection.

A characteristic effect of SAgAs was the dramatic upregulation of CD73 on both adoptively transferred and endogenous antigen-specific T cells, particularly with cSAgAs, whereas this upregulation was not induced by free peptides. CD73 is an ectonucleotidase that, in concert with CD39, generates extracellular adenosine, an immunosuppressive metabolite (26). Upregulation of CD73 on T cells has been associated with disease protection in NOD mice (22), and T cells that coexpress CD73 and FR4 have previously been defined as anergic (21). Interestingly, it has been reported that such T cells may become regulatory if they also express Nrp-1 (21,27); therefore, Nrp-1 would be an important marker to follow in future SAgA studies. Furthermore, KLRG1 was found to be induced on both p79- and 2.5HIP-reactive CD4⁺ T cells and PD-1 only on the p79-reactive ones (possibly requiring stronger stimulation). KLRG1 has recently emerged as a new checkpoint molecule for T cells and NK cells (28–30), and both KLRG-1 and PD-1 have been associated with anergic and exhausted-like T-cell phenotypes achieved during immunotherapy in patients with T1D (31–33). Our studies demonstrate the potential of SAgAs to improve delivery of self-peptides and the tolerogenic engagement of autoreactive T cells, potentially resulting in anergy and/or exhaustion of specific T cells. Because persistent antigen exposure may promote exhaustion, the prolonged antigen presentation observed after SAgA treatment may contribute to this result. If antigen presentation lasts close to a week, weekly SAgA treatment may indeed result in persistent antigen exposure.

Although both peptides stimulated BDC2.5 T cells, the responses of endogenous p79- and 2.5HIP-reactive T cells were largely nonoverlapping in staining and reactivity. Indeed, MHC-II Tetrs loaded with these two peptides identified discreet populations of CD4⁺ T cells, and to our surprise, few cells were stained with both Tetrs, and cross-reactivity in T-cell responses was negligible. Moreover, SAgA_p79 or SAgA_2.5HIP elicited different responses. The p79 mimotope is the most potent synthetic peptide, identified from a library of artificial peptides, that stimulates BDC2.5 T cells (22). In contrast, 2.5HIP is thought to be the most potent natural peptide for BDC2.5 T cells identified to date (14). Delivery of the same 2.5HIP peptide using nanoparticles was recently shown to efficiently prevent disease in NOD.SCID mice transferred with BDC2.5 T cells (monoclonal model of disease development) (21). Multiple administrations of SAgA_p79 led to substantial expansion of p79-reactive T cells, with a high proportion of CD73⁺ FR4⁺ cells and IL-10⁺ Tr1 cells predominating over IFN-γ⁺ Th1 cells. In contrast, the phenotype of 2.5HIP-reactive T cells remained largely unchanged after two doses, with a predominance of IFN-γ production over IL-10. The induced KLRG1⁺ cells were primarily IFN-γ⁺ in 2.5HIP-reactive T cells and IFN-γ⁻ in p79-reactive T cells. There are multiple possible explanations for these disparate responses. First, when delivered as SAgAs, the p79 epitope is ∼10 times more active on BDC2.5 T cells than 2.5HIP in vitro. Second, the 2.5HIP peptide has lower water solubility than p79 (as determined by the PepCalc.com calculator), which can influence overall SAgA solubility and its biodistribution in vivo. Indeed, our data showing reduced uptake of FITC-SAgA_2.5HIP may reflect more limited in vivo diffusion because of the difference in solubility. Third, 2.5HIP is a natural antigen in NOD mice; therefore, 2.5HIP-reactive T cells may have been more antigen experienced than p79-reactive T cells, which may explain the higher initial frequency of IFN-γ⁺ cells among 2.5HIP-reactive T cells. Thus, the engagement of specific T cells to peptides from SAgA_p79 and SAgA_2.5HIP may differ as a result of peptide properties (e.g., MHC binding affinity and T-cell receptor reactivity), solubility/dispersion properties affecting local antigen dose, and prior antigen exposure. This differential response to two peptides is interesting and highlights the importance of assessing the mechanisms of action over multiple antigens for the same delivery approach.

A differentiating feature of SAgAs is the apparent complementarity of SAgA_p79 and SAgA_2.5HIP (at 2.5 nmol each) in arresting disease development in NOD mice. Treatment with SAgA_p79 or SAgA_2.5HIP alone (2.5 nmol) had no significant effect on disease incidence. Even the mixture at 0.5 nmol each had an effect that was at least as good as individual hSAgAs at 2.5 nmol, suggesting that it is not a dose issue and that the two SAgAs may indeed synergize. The strong protection conferred by SAgA_mix (with these two peptides) suggests that active suppression is likely involved through the complementary action of IL-10⁺ Tr1 cells (induced by p79) and possibly Foxp3⁺ Tregs (induced by the less stimulatory 2.5HIP) to regulate the other diabetogenic T-cell clones. Moreover, bystander suppression may also be mediated via CD73 by promoting a more immunosuppressive local environment with adenosine. Treatment with SAgAs did not lead to deletion of T cells reactive to both SAgA-linked peptides and other β-cell antigen-related epitopes. On the contrary, the repeated delivery of SAgAs substantially expanded antigen-specific T cells, and although their frequency dwindled over time, they remained detectable several months after cessation of treatment. This maintained population of antigen-specific T cells may be needed to regulate other T-cell populations. Clonal expansion and anergy are not incompatible and rather reflect two consecutive stages in the T-cell response over time (34,35). This study provides important insights into how different peptides with similar sequences, same delivery method, and same dose can have remarkably different effects on specific T cells in vivo.

Our studies demonstrate that 2.5HIP and p79, delivered as SAgAs, stimulate nonoverlapping endogenous T-cell populations and significantly block spontaneous disease development in NOD mice. The two epitopes may be synergistic in combination, and delivery of more epitopes may further enhance the efficacy of this ASIT approach. Anaphylactic reaction to administered peptides, including natural islet peptides, has been reported in NOD mice (2,3). Delivery as SAgAs could attenuate or negate such adverse effects that some peptides may elicit, and this novel property of SAgAs will be further investigated, given the important implications in peptide-based ASIT. Because the free peptides most strongly induced anaphylaxis, we posit that cSAgAs, with more stable linkers, are less likely to shed free peptides before uptake. When translating this approach to patients, epitopes may be customized based on their MHC haplotypes, autoantibodies, and other immune markers and delivered in combination to enhance the efficacy of this ASIT approach. Our studies support the use of mimotopes and hybrid peptides to target disease-relevant T cells and possibly complement more conventional antigens. Thus, the SAgA platform is expected to enhance the delivery, efficacy, and possibly safety of peptide-based immunotherapy for the treatment of autoimmune diabetes.

Acknowledgments. The authors thank Sandy Khong, Dr. Randy Pi, Dr. Franziska Matzkies, Francis Campana, and Adrien Lanusse for additional support. The authors thank the National Institutes of Health Tetramer Core Facility, supported by contract HHSN272201300006C from the National Institute of Allergy and Infectious Diseases, for the MHC Tetr provided for these studies.

Funding. R.F.-F. was supported by Postdoctoral Minority Fellowship Award 1-19-PMF-022 from the American Diabetes Association. S.N.J. and M.A.L. were supported by National Institutes of Health Graduate Training Program in Dynamic Aspects of Chemical Biology grant T32GM008545 from the National Institute of General Medical Sciences. These studies were funded by JDRF grant 2-SRA-2017-312-S-B to J.O.S. Research reported in this publication was performed using the CCTI Flow Cytometry Core, supported in part by the Office of the Director, National Institutes of Health, under award S10OD020056 and National Center for Research Resources award S10RR027050 and by Diabetes Research Center grant P30DK063608.

Duality of Interest. J.O.S. and C.B. are co-founders and shareholders of Orion BioScience, Inc., which is developing SAgA technology for treating autoimmune diseases.

Author Contributions. R.F.-F. and R.J.C. designed experiments and wrote the first draft of the manuscript. R.F.-F. performed experiments and analyzed data. S.N.J., M.A.L., J.O.S., and C.B. produced and characterized SAgAs. M.K.-M. assisted with tissue isolation and analysis. R.L.B. validated and provided 2.5HIP-Tetr. R.J.C. directed the research. R.F.-F., S.N.J., R.L.B., J.O.S., C.B., and R.J.C. edited the manuscript. All authors approved the manuscript. R.F.-F. 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. This study was presented in part at the 2019 Federation of Clinical Immunology Societies (FOCIS) Annual Meeting, Boston, MA, 18–21 June 2019 and the 2020 FOCIS Virtual Annual Meeting, 28–31 October 2020.