Type 1 diabetes (T1D) is an autoimmune disease characterized by T and B cell responses to proteins expressed by insulin-producing pancreatic β cells, inflammatory lesions within islets (insulitis), and β cell loss. We previously showed that Ag-specific tolerance targeting single β cell protein epitopes is effective in preventing T1D induced by transfer of monospecific diabetogenic CD4 and CD8 transgenic T cells to NOD.scid mice. However, tolerance induction to individual diabetogenic proteins, for example, GAD65 (glutamic acid decarboxylase 65) or insulin, has failed to ameliorate T1D both in wild-type NOD mice and in the clinic. Initiation and progression of T1D is likely due to activation of T cells specific for multiple diabetogenic epitopes. To test this hypothesis, recombinant insulin, GAD65, and chromogranin A proteins were encapsulated within poly(d,l-lactic-co-glycolic acid) (PLGA) nanoparticles (COUR CNPs) to assess regulatory T cell induction, inhibition of Ag-specific T cell responses, and blockade of T1D induction/progression in NOD mice. Whereas treatment of NOD mice with CNPs containing a single protein inhibited the corresponding Ag-specific T cell response, inhibition of overt T1D development only occurred when all three diabetogenic proteins were included within the CNPs (CNP-T1D). Blockade of T1D following CNP-T1D tolerization was characterized by regulatory T cell induction and a significant decrease in both peri-insulitis and immune cell infiltration into pancreatic islets. As we have recently published that CNP treatment is both safe and induced Ag-specific tolerance in a phase 1/2a celiac disease clinical trial, Ag-specific tolerance induced by nanoparticles encapsulating multiple diabetogenic proteins is a promising approach to T1D treatment.
This article is featured in Top Reads, p.
Type 1 diabetes (T1D) is an autoimmune disease affecting ∼1.6 million people in the United States and ∼4.7 million people world-wide (1, 2). Presently, the standard of care for patients with T1D is exogenous insulin replacement combined with diet and exercise modifications (3). Evidence from both the NOD mouse model of spontaneous T1D and T1D patients suggests that insulin is only one of several dominant measurable disease-associated Ags recognized by autoreactive CD4+ T cells and CD8+ T cells driving disease pathogenesis (4, 5). The disease initiating epitope or epitopes in T1D are not known, and therefore, not surprisingly, single protein or peptide tolerance approaches have been unsuccessful to date in both NOD mice and human T1D patients. T1D is not a single Ag-mediated disease, rather once inflammation within the pancreas and tissue destruction occur, the autoimmune response expands to include both T cell and B cell responses to additional epitopes, a process known as epitope spreading (6–9). In support of a role for epitope spreading in T1D, autoreactive responses to Ags other than insulin, including glutamic acid decarboxylase 65 (GAD65), islet-specific glucose-6-phosphate catalytic subunit-related protein (IGRP), islet Ag 2 (IA-2), phogrin (IA-2β), chromogranin A, zinc transporter 8 (ZnT8), and vasostain-1 (5, 10–12), have been identified in both T1D patients and NOD mice. Therefore, the failure of tolerance approaches targeting single Ags, such as GAD65 or insulin, to inhibit the onset and eventual progression of disease in clinical trials suggests that tolerance to multiple disease Ags may be necessary to provide efficient disease control (6, 8, 9).
Currently, there are no available therapies that treat the root cause of T1D, that is, activated autoimmune T cells mediating continued pancreatic β cell destruction, either through CD8+ T cells, CD4+ T cell–mediated inflammation, or autoantibodies. Therapies in the pipeline are alternatively targeting broad T cell suppression with anti-CD3 or transient suppression with single Ag-specific chimeric Ag receptor regulatory T (Treg) cells. Our laboratory has studied the ability to induce Ag-specific tolerance via the induction of Ag-specific Treg cells and the induction of IL-10, CTLA-4, and PD-1 expression following treatment with Ag-coupled apoptotic splenocytes, Ag-coupled nanoparticles, and nanoparticles that contain encapsulated Ag (reviewed in Refs. 13, 14). We have shown that i.v. treatment with carboxylated poly(d,l-lactide-co-glycolide) (PLGA) nanoparticles induces Ag-specific immune tolerance in a wide variety of mouse disease models via CD4+ Treg cell–dependent mechanisms (3, 15–17). Biodegradable carboxylated PLGA nanoparticles containing encapsulated Ag (tolerance-inducing microparticles [TIMPs]; named COUR nanoparticles [CNPs]) were shown to prevent and treat disease in several mouse models of T1D through reprogramming of APCs and induction of Treg cells. Nanoparticle size (400–800 nm) and negative charge results in phagocytic uptake of the Ag-encapsulating nanoparticles by tolerogenic APCs via the macrophage receptor with collagenous structure (MARCO) scavenger receptor in the splenic marginal zone and liver, resulting in presentation of the Ag in the context of immune regulatory signals including PD-L1/2 and IL-10 (14). For example, the treatment with PLGA nanoparticles encapsulating or coated with the chromogranin A mimotope peptide, p31, following the transfer of diabetogenic transgenic CD4+ BDC-2.5 T cells to NOD.scid mice, led to the sequestration of the autoreactive effector CD4+ T cells within the spleen and increased CD4+Foxp3+ Treg cells in the spleen, pancreatic lymph nodes, and pancreas (16). The induction of the Treg cells was found to be required for long-term tolerance, as the removal of CD25+ T cells drastically reduced the time through which tolerance was maintained. Additionally, in the BDC-2.5 CD4+ T cell adoptive transfer model, PLGA nanoparticles coupled with a hybrid insulin peptide, 2.5HIP, also increased the numbers of Foxp3+ Treg cells in the pancreas and reduced the trafficking of effector CD4+ T cells (18). These findings suggest that a common regulatory mechanism is induced at the cellular level across mouse strains, in that Ag-loaded PLGA nanoparticle treatment induced various subsets of Ag-specific Treg cells capable of regulating autoimmune disease.
With regard to clinical translation, CNPs have demonstrated efficacy to induce tolerance in mouse models of experimental autoimmune encephalomyelitis (EAE), T1D (adoptive transfer), celiac disease, and Th2-induced allergic airway inflammation (16, 17, 19). In these models, CNP treatment reduced pathogenic cytokine production (IFN-γ, IL-17, IL-4, and IL-5) (16, 17, 19), increasing anti-inflammatory IL-10 cytokine production (16) and increasing CD25+Foxp3+ Treg cells at the site of disease (16). In preclinical studies to assess CNP function in celiac disease, treatment with gliadin encapsulating PLGA nanoparticles (TIMP-GLIA, TAK-101/CNP-101) significantly decreased gliadin-specific T cell proliferation, inflammatory cytokine secretion, circulating gliadin-specific IgG/IgG2c, gliadin-specific DTH response, gluten-dependent enteropathy, and weight loss (17). The functionality of TAK-101/TIMP-GLIA treatment was also demonstrated in HLA-DQ8 transgenic mice, as treatment increased the level of Foxp3 expression and induced gene signatures associated with tolerance induction (17). The CNP design was first introduced into human clinical studies with TAK-101/TIMP-GLIA and demonstrated successful induction of tolerance in celiac patients orally challenged with gluten (20). These studies assessed safety, tolerability, pharmacokinetics, and efficacy of TAK-101/TIMP-GLIA treatment to induce gliadin-specific T cell tolerance in patients with biopsy-confirmed celiac disease (20).
In the current study we sought to address both a fundamental immunological question regarding identification of disease-dependent Ag-specific T cell responses involved in the onset of T1D in NOD mice and to determine whether treatment of NOD mice with CNPs containing recombinant diabetogenic proteins would inhibit the development of T1D. The spontaneous NOD mouse model recapitulates some key disease characteristics of human T1D, where multiple diabetogenic Ags drive disease progression. Insulin, GAD65, and chromogranin A proteins known to be potential targets in TID were encapsulated within TIMPs (CNP-T1D), enabling us to test the simultaneous tolerogenic targeting of multiple diabetogenic CD4+ T cell and CD8+ T cell epitopes for efficacy in preventing T1D onset (21). Treatment with CNP-T1D beginning at either 6 or 10 wk of age prevented β cell destruction and further epitope spreading, as well as inhibiting the secretion of IFN-γ upon splenocyte ex vivo recall culture. Encapsulation of multiple disease epitopes in CNP-T1D covering three major T1D autoantigens was necessary to prevent T1D onset, as single protein tolerance proved insufficient. CNP-T1D–induced tolerance was durable, lasting for 13.5 wk after third dose administration. Taken together, these data show that CNP-T1D is a promising therapeutic for treatment of T1D and provides an advantage over single protein or single epitope targeting therapies by inducing tolerance to multiple encapsulated cognate epitopes as well as spread epitopes.
Materials and Methods
Acid-terminated PLGA polymer was purchased from Lactel absorbable polymers (Evonik, Mobile, AL). Recombinant insulin was purchased from Sigma-Aldrich (St. Louis, MO). Recombinant GAD65 and chromogranin proteins were custom ordered from Biomatik (Kitchener, ON, Canada). Myelin oligodendrocyte glycoprotein (MOG)92–106, p31, Proins C19-A3, and the linkered peptide p31-NRPA7-InsB9–23 were purchased from Genemed Synthesis (Torrance, CA). 2.5-HIP was generously provided by the Kathryn Haskins laboratory at the University of Colorado. See Supplemental Table I for peptide sequences.
Manufacture of CNPs (PLGA nanoparticles)
For CNPs encapsulating diabetic proteins (insulin, chromogranin A, or GAD65), PLGA solution was mixed with each recombinant protein to generate a water-in-oil emulsion. This was mixed with a proprietary blend of surfactants and organic solvents to form an oil-in-water secondary emulsion. The solvent was removed by evaporation, yielding PLGA nanoparticles encapsulating diabetic protein, which were washed, filtered, and concentrated via tangential flow filtration. CNP-insulin, CNP–chromogranin A, and CNP-GAD65 were supplied as a lyophilized powder containing ∼100 mg of PLGA nanoparticles per vial with average nanoparticle diameters 400–800 nm and ζ potentials of −30 to −60 mV. The protein content was ≥3 μg of recombinant protein per milligram of PLGA (Supplemental Fig. 1). Nanoparticles used in initial attempts to inhibit T1D were created by slightly different methods. PLGA(insulin) was created by double-emulsion, solvent evaporation as described previously (22). PLGA/MOG92–106, PLGA/insulin, PLGA/Proins C19-A3, and PLGA/p31-NRPA7-InsB9–23 were synthesized by first coupling peptide to PLGA polymer in DMSO using ethylene carbodiimide, followed by single-emulsion solvent evaporation as previously described (23).
NOD mice, blood glucose monitoring, and treatment
Female NOD mice, 5 wk of age, were purchased from The Jackson Laboratory (Bar Harbor, ME). All mice were housed under specific pathogen-free conditions in the Northwestern University Center for Comparative Medicine and maintained according to protocols approved by the Northwestern University Animal Care and Use Committee (Chicago, IL). NOD mice were treated with CNPs (2.5 mg/dose in 200 μl of sterile saline) on the specified days. Blood glucose levels were measured in female NOD mice with the UltraTRAK PRO blood glucose monitoring system weekly starting at 6 wk of age. Mice with two consecutive readings ≥250 mg/dl were determined to be diabetic.
Ex vivo recalls
On indicated days mice were euthanized and spleens were collected for isolation of single-cell suspensions. For ex vivo reactivation cultures, 1 × 106 total splenocytes were plated in replicate wells on 96-well flat-bottom plates with complete RPMI 1640 medium plus anti-CD3 (1 μg/ml), OVA323–339, chromogranin A359–372, insulin B9–23, insulin A14–20, GAD65286–300, GAD65515–524, IGRP206–214,heat shock protein (HSP)60437–460, or Znt8345–359 (20 μg/ml) (Supplemental Table II). The culture supernatants were collected at 72 h after culture and levels of secreted cytokines were determined via Luminex using a mouse cytokine/chemokine magnetic bead panel (Millipore).
Pathology scoring of pancreata
For histological analysis, the pancreas was removed and snap-frozen in OCT. Multiple 10 μm sections were stained with H&E and scored blindly for insulitis (score: no infiltrate; peri-insulitis present; <25%; and >25% of the islet is infiltrated). Average insulitis percentages were determined from the total number of islets counted from each treatment group.
Tissue processing and flow cytometry
The spleens and pancreatic lymph nodes were processed via physical disruption, and RBCs were lysed using ammonium chloride prior to processing for flow cytometry. For the pancreas, single-cell suspensions were prepared by mincing the tumor tissue in 2 ml of Accutase (Millipore) plus 1 mg/ml collagenase, and the samples were incubated at 37°C for 30 min. Following enzyme digestion, the pancreas samples were disrupted with a 100-μm cell strainer, the cell strainer washed twice with 10 ml of HBSS + 5% FCS, and the cells were pelleted. Cells were washed in PBS, stained with LIVE/DEAD fixable aqua dead cell stain (Life Technologies, Grand Island, NY), blocked with anti-CD16/32 (Thermo Fisher Scientific), and then stained with the indicated Abs. Viable cells (106) were analyzed per individual sample using a BD FACSCelesta (BD Biosciences), and the data were analyzed using FlowJo version 9.5.2 software (Tree Star, Ashland, OR). The specific Abs used are presented in Supplemental Table III.
Comparisons of T1D incidence were analyzed by χ2 using Fisher’s exact probability. Two-way ANOVA with a Bonferroni posttest was used to determine statistical differences between treatment groups.
Multiple diabetogenic peptide-specific T cell populations are present within the spleen of NOD mice
Identification of the specific disease-inducing protein/peptide in spontaneous T1D in NOD mice and human patients has been a topic of study for many years. Although the induction of tolerance to a single peptide is very reproducible after i.v. infusion of peptide-encapsulating PLGA nanoparticles (TIMPs; named CNPs), the specific disease-associated T cell epitopes need to be known for this tolerance therapy to be effective. We previously showed that treatment of NOD.scid mouse PLGA nanoparticles encapsulating with the BDC2.5 mimotope 1040-31 (p31), the NY8.3 mimotope NRPA7, and the linked p31-NRPA7-insulin B9–23 peptide significantly inhibit T1D-induced by the transfer of transgenic CD4+ BDC2.5 or CD8+ NY8.3 T cells, respectively (16). However, individual treatment of wild-type NOD mice with PLGA nanoparticles containing p31, p31-NRPA7-insulin B9–23, Proins C19-A3, or intact insulin (Fig. 1A) failed to inhibit development of T1D. Similarly, although treatment of NOD.scid mice alone with PLGA-2.5HIP inhibits CD4+ BDC2.5 transgenic T cell–induced T1D (18), this treatment is not able to delay the onset of T1D in NOD mice (Fig. 1B). Thus, targeting of individual diabetogenic peptide epitopes is incapable of inhibiting onset of T1D, supporting the hypothesis that tolerogenic targeting of multiple CD4+ T cell and CD8+ T cell epitopes using a combination of several recombinant diabetogenic proteins may be required.
Based on previously published data and the above results, we asked whether detectible T cell responses to multiple diabetogenic peptides were present within the spleen of 6 and 9 wk-old NOD mice. These two time points were chosen as this represents ages where the number of immune cells within the pancreatic islets increases (24), there is detectible circulating anti-insulin Ab (25), and there is a detectable splenic T cell response to various insulin peptides (26, 27). Splenocytes from 6- and 9-wk-old NOD mice and 6-wk-old C57BL/6 mice were cultured in the presence of 20 μg/ml of the following peptides: OVA323–339, chromogranin A359–372, insulin B9–23, insulin A14–29, GAD65286–300, GAD65515–524, IGRP206–214, HSP60437–460, or ZnT8345–359 (20). At 72 h of culture, supernatants were collected, and the levels of secreted IFN-γ were measured. We found that splenocytes from 6-wk-old NOD mice, but not C57BL/6 mice, produced low, but significant, levels of IFN-γ in response to all of the diabetogenic peptides, but not to the control OVA323–339 peptide (Fig. 1C). Additionally, there is a trend toward an age-dependent increase in the level of IFN-γ secreted by splenocytes cultured in the presence of each of the diabetogenic peptides, with the interesting exception of GAD65515–524. These findings indicate that T cells of various specificities are present and activated within young NOD mice, as early as 6 wk of age, well prior to the onset of overt T1D.
CNP-T1D treatment inhibits the development of T1D in NOD mice
Based on the above data, we tested whether the induction of tolerance to multiple diabetogenic proteins would block the onset of T1D. NOD mice were randomized into two treatment groups and treated with either 2.5 mg/dose of unloaded CNP or CNPs containing recombinant insulin, GAD65, and chromogranin A proteins. A 2.5 mg dose of CNP-T1D consisted of 0.83 mg of CNP-insulin, 0.83 mg of CNP-GAD65, and 0.83 mg of CNP–chromogranin A, equaling a total 2.5 mg/dose. We hypothesized that CNP-T1D treatment would induce tolerance to chromogranin A359–372, insulin B9–23, insulin A14–29, GAD65286–300, and GAD65515–524, and thereby allow for an enhanced regulatory response within both the spleen and pancreas. At 6 wk of age, NOD mice have anti-insulin Abs (26), which are associated with T1D progression and are similar to the Ab profile for prediabetic stage 2 T1D patients (27). Studies in NOD mice (28) and humans (29–30) suggest that 40–85% of the β cell mass has been destroyed by the onset of diabetes (stage 3 in humans). Therefore, tolerogenic therapies administered prior to complete destruction of the β cell mass may serve the greatest benefit for patients. Mice were treated with 2.5 mg/dose of unloaded CNP or CNP-T1D at 6, 7, and 11 wk of age to replicate stage 2 disease in T1D patients, and the level of blood glucose was measured biweekly. Mice were considered to have overt diabetes after two consecutive blood glucose readings of >250 mg/dl. As shown in (Fig. 2A, treatment with CNP-T1D significantly decreased the number of NOD mice that developed T1D (1/10) versus CNP alone (9/10), and the effect was durable for at least 13.5 wk after the third dose. CNP-T1D treatment significantly decreased the percentage of pancreatic islets that contained >25% immune cells (Fig. 2B, 2C). Approximately 50% of the islets within the pancreas of CNP-T1D–treated NOD mice had no detectable infiltrating immune cells. Additionally, the greatest efficacy of CNP-T1D required two or three doses (Fig. 2D).
To better simulate treatment closer to the onset of dysglycemia, we next tested CNP-T1D treatment of NOD mice at 10, 11, and 15 wk of age. Similar to mice treated at 6, 7, and 11 wk of age, this delayed treatment regimen significantly decreased the incidence of mice developing T1D (Fig. 2E) as well as the percentage of pancreatic islets that contained >25% immune cells (Fig. 2F), and the effect was durable for at least 7 wk after third dose administration. Collectively, these results show that tolerogenic treatment with CNP-T1D containing multiple diabetogenic CD4+ T cell and CD8+ T cell epitopes starting at either 6 wk of age to mimic stage 2 T1D (prior to onset of clinical symptoms) or at 10 wk of age to mimic late stage 2/early stage 3 T1D (prior to overt diabetes but after dysglycemia) significantly prevented T1D development, which developed in >90% of control mice starting at 13 wk of age.
Durable tolerance requires all three diabetogenic proteins in CNP-T1D and decreases IFN-γ and increases IL-10 secretion
As further confirmation for the requirement of all three diabetogenic proteins to be contained within the CNP-T1D dose, we compared tolerance with the individual recombinant proteins to CNP-T1D. NOD mice were treated with unloaded CNP (2.5 mg/dose), CNP-insulin (0.83 mg/dose + 1.67 mg/dose of unloaded CNP), CNP–chromogranin A (0.83 mg/dose + 1.67 mg/dose of unloaded CNP), CNP-GAD65 (0.83 mg/dose + 1.67 mg/dose of unloaded CNP), or CNP-T1D (2.5 mg/dose) at 6, 7, and 11 wk of age. As such, all mice received the same dose of 2.5 mg of CNPs, and the amount of each respective recombinant diabetogenic protein within this dose was constant across respective treatment groups. Strikingly, the results show that only treatment of NOD mice with CNP-T1D containing the three diabetogenic proteins significantly inhibited the onset of T1D as compared with nanoparticles encapsulating any one of the individual recombinant proteins (Fig. 3A). In the present experiment, NOD mice treated with CNP-insulin and CNP–chromogranin A exhibited a significant increase in disease incidence as compared with the unloaded CNP treatment group (Fig. 3A). However, this finding varied from experiment to experiment, as PLG(insulin) and PLG/insulin did not increase in disease incidence, as presented in (Fig. 1A.
To confirm the specificity of the various tolerogenic treatments, IFN-γ and IL-10 production were measured from splenocyte ex vivo recall cultures at 9 wk of age, a time point when the mice had received two doses of the indicated CNPs. Total splenocytes from individual mice were cultured in the presence of OVA323–339, chromogranin A359–372, insulin A14–29, insulin B9–23, GAD65286–300, GAD65515–524, IGRP206–214, HSP60437–460, or ZnT8345–359 (20 μg/ml) for 72 h (Fig. 3B–J). The treatment of NOD mice with CNP-T1D significantly decreased the level of IFN-γ secreted upon splenocyte culture in the presence of insulin A14–20, insulin B9–23, and IGRP437–460, while also increasing the level of IL-10 secreted upon splenocyte culture in the presence of insulin A14–20, insulin B9–23, GAD65286–300, GAD65515–524, and chromogranin A359–372 (Fig. 3C–H). In contrast, CNP-insulin only decreased the level of IFN-γ secreted upon splenocyte culture in the presence of insulin A14–20, insulin B9–23, and IGRP437–460. Although CNP-GAD65 treatment increased the level of IL-10 secreted upon splenocyte culture similar to CNP-T1D, CNP-GAD65 treatment did not significantly modulate the level of IFN-γ secreted (Fig. 3E). Lastly, CNP–chromogranin A significantly decreased the level of IFN-γ secreted upon splenocyte culture in the presence of chromogranin A359–372 and IGRP437–460 (Fig. 3G, 3H). Therefore, splenocytes from CNP-T1D–treated mice additively presented the functional changes for splenocytes from mice that had been treated with CNP-insulin, CNP-GAD65, or CNP–chromogranin A alone. These findings indicate that CNP-T1D induced the broadest overall tolerogenic phenotype in the self-reactive T cells, and this modulation of the Ag-specific T cell response to T1D-asscoaited peptides correlates with the disease course data (Fig. 3A).
CNP-T1D induces a regulatory phenotype within the spleen, pancreatic lymph nodes, and pancreas
To confirm the ex vivo findings following treatment with CNP-T1D versus treatment with the individual proteins alone, splenocytes were analyzed by flow cytometry (Fig. 4A). CNP-T1D or CNP–chromogranin A treatment significantly increased the number of total CD4+ T cells, macrophages, and monocytes increased within the spleen, as compared with unloaded CNP-treated mice. When comparing the phenotype of the CD4+ T cells, macrophages, and monocytes within the spleen, CNP-T1D treatment significantly increased the number of Foxp3+ and PD-1+ CD4+ T cells (Fig. 4B, 4C), as well as increasing the number of PD-L1+ and CD206+ macrophages and PD-L1+ monocytes (Fig. 4B, 4E, 4F). In contrast, CNP-GAD65 treatment, which also increased the level of secreted IL-10 similar to CNP-T1D treatment (Fig. 3C–G), increased the number of IFN-γ+ and IL-17+ CD4+ T cells within the spleen (Fig. 4C). Additionally, the finding that CNP-GAD65 and CNP-T1D induced an increase in the level of IL-10 secreted, while not significantly increasing the number of IL-10+CD4+ T cells within the spleen, suggests that treatment induced an increase in the amount of IL-10 secreted per IL-10+CD4+ T cell at the 9 wk of age time point. These present data correlate with our previously published data in the BDC2.5 transfer disease into NOD.scid mice (16). Taken together, the data show that although mice treated with the individual diabetogenic protein containing CNPs may induce some of the same regulatory alterations as CNP-T1D, it is only the mice that were treated with CNP-T1D that have the combined effect of decreasing IFN-γ and increasing IL-10.
To further assess the immune cell population differences, flow cytometry analyses of the spleen, pancreatic lymph nodes, and pancreas were carried out at a later stage of disease (12 wk of age, when controls begin to develop overt T1D) in NOD mice that had been treated with unloaded CNP, CNP-T1D, or the individual recombinant diabetogenic proteins at 6, 7, and 11 wk of age. CNP-T1D treatment increased the number of total immune cells within the spleen while significantly decreasing the number of total immune cells within both the pancreatic lymph nodes and the pancreas (Fig. 5A). When compared with the individual proteins, only CNP-T1D treatment significantly increased the number of CD4+ T cells, macrophages, and monocytes within the spleen (Fig. 5B). The same pattern was found at 9 wk of age, which is after only two doses of CNP-T1D (Fig. 4B). Additionally, CNP-T1D significantly decreased the number of CD4+ T cells and CD8+ T cells within the pancreatic lymph nodes and pancreas (Fig. 5C, 5D). These findings indicate a shift of immune cells away from the disease-associated target tissue, that is, the pancreas.
We also assessed the phenotype of the CD4+ T cells in the various organs. The percentage of Foxp3+, CTLA-4+, and PD-1+ CD4+ T cells increased in both the spleen and pancreas of CNP-T1D–treated mice (Fig. 5E–G). The present data show that CNP-GAD65 and CNP-T1D treatment increased the percentage of IL-10+ Treg cells within the spleen, pancreatic lymph nodes, and pancreas, whereas CNP–chromogranin A only increased the percentage within the pancreas (Fig. 5E–G, Supplemental Fig. 2). When considered with the CNP-T1D–induced decrease in the number of total immune cells and CD4+ T cells within the pancreas as compared with the unloaded CNP-treated NOD mice, this equates to a significant decrease in the number of IFN-γ+ and IL-17+ CD4+ T cells present within the pancreata along with an increase in Treg cells. This finding is consistent with the histopathology scoring data presented in (Fig. 2C and 2F. Taken together, the present findings suggest that CNP-T1D treatment induces a regulatory phenotype in an Ag-specific manner for multiple T cell specificities, and thereby inhibits the inflammatory immune response within the pancreas.
Previous studies have attempted tolerogenic treatment of T1D by targeting individual diabetogenic Ags, with limited success (8, 31–34). Indeed, our attempts to prevent/delay onset of T1D in wild-type NOD mice using PLGA nanoparticles coupled with or encapsulating a number of dominant natural and hybrid diabetogenic peptide epitopes (Fig. 1A, 1B) were unsuccessful even though these formulations were previously shown to robustly inhibit monospecific T1D induced by transfer of transgenic CD4+ T cells and CD8+ T cells into NOD.scid recipients (16, 18, 35). It is well established that the genesis of T1D in humans and NOD mice is characterized by the presence of autoreactive T cell and Ab responses to multiple pancreatic β cell Ags that arise well before the onset of overt hyperglycemia (5, 25–27) likely via epitope spreading (6–9). Thus, to determine whether simultaneous tolerogenic targeting of multiple CD4+ T cell and CD8+ T cell diabetogenic epitopes may be required to block development of T1D in NOD mice, we assessed disease in mice treated with PLGA nanoparticles encapsulating either recombinant insulin, GAD65, or chromogranin A, alone or in combination (CNP-T1D), beginning at 6 wk of age. Indeed, disease was only blocked in mice receiving CNP-T1D, but not in mice receiving nanoparticles encapsulating any of the individual diabetogenic proteins (Figs. 2A–C, 3A). This indicates that simultaneous targeting of multiple autoepitopes is required for efficient disease protection in a multideterminant autoimmune disease to allow development of a critical mass of regulatory responses. Furthermore, effective tolerance required two to three doses of CNP-T1D administered at 6, 7, and 12 wk of age (Fig. 2D).
In support of the requirement for multiple diabetogenic peptides to be contained within the Ag-specific tolerance treatment, for example, CNP-T1D, we detected low, but significant and reproducible, proinflammatory IFN-γ production by NOD T cells in response to at least eight different diabetogenic peptide epitopes within six different β cell proteins (chromogranin, insulin, GAD65, IGRP, HSP60, and ZnT8) by 6 wk of age (Fig. 1C), a time point well prior to development of overt T1D, which onsets between 13 and 16 wk of age in our vivarium. Additionally, responses to seven of the eight peptides show a trend toward an increase by 9 wk of age, and no responses were seen in 6-wk-old T1D-resistant C57BL/6 mice. The breadth of β cell autoepitopes targeted in NOD mice is perhaps not surprising, as epitope spreading has been previously well documented in various experimental autoimmune disease models, including EAE (36) and T1D in NOD mice (37, 38). Epitope spreading can occur to different epitopes within the same self-antigen (intramolecular epitope spreading) or to epitopes on a different self-antigen (intermolecular epitope spreading) and has critical implications to tolerance-based immunotherapy, as the effectiveness would require induction of unresponsiveness to multiple autoepitopes in pre-established disease (7). Relevant to disease therapy and the location of initial autoreactive CD4+ T cell activation, we have shown in relapsing EAE induced by priming SJL/J mice with PLP178–191/CFA that naive CFSE-labeled PLP139–151-specific transgenic T cells that drive the primary relapse (intramolecular epitope spreading) are initially activated directly within the CNS target organ and not in the draining cervical lymph nodes or other peripheral lymphoid organs (39). The spread epitope-specific T cells are activated by peripherally derived myeloid dendritic cells that infiltrate the CNS during the acute disease episode (40, 41). Thus, epitope spreading takes place in a hierarchical order of peptide dominance, and activation of the spread epitope-specific CD4+ T cells occurs within the inflamed target organ and can be inhibited by tolerance induction to the spread epitope (23). Therefore, blockade of disease-associated tissue inflammation and destruction by Ag-specific tolerance may also inhibit the activation of spread epitope-specific T cells that are responsive to peptides not contained within CNP-T1D. We have also shown that early life oligodendrocyte ablation using diphtheria toxin A (DTA) in the C57BL/6 DTA mouse model (42) leads to a later life fatal, secondary demyelinating disease driven by MOG-specific T cells developing ∼30 wk after recovering from initial oligodendrocyte loss and demyelination. Importantly, MOG35–55-specific tolerance induced by MOG35–55-encapsulating PLGA nanoparticles treatment blocks the progression of this late-onset disease (43). Thus, Ag-specific CNP treatment can function as a potent therapeutic treatment in chronic autoimmunity induced by either inflammatory T cells (EAE) or secondary to noninflammatory cell death (DTA mice) in the target organ.
In addition to assessing the Ag specificity of CNP-T1D treatment, our studies demonstrate the durability of CNP-T1D–induced tolerance. Mice treated with CNP-T1D at 6, 7, and 11 wk of age remained disease free through 25 wk of age, that is, at least 13.5 wk after third dose administration (Fig. 2A). Tolerance durability was assessed by the level of blood glucose remaining <250 mg/dl and a marked reduction in inflammatory cell infiltration into the pancreas leading to preserved islet architecture. We have previously demonstrated the durability of nanoparticle-induced Ag-specific tolerance and its ability to halt previously established autoimmune disease in myelin peptide/CFA-induced EAE models utilizing a single peptide to induce CD4+ T cell–mediated autoimmunity (15, 23). In addition to the ability of CNP-T1D to protect from T1D development in NOD mice treated at 6 wk of age, we notably show similar protection in NOD mice treated beginning at 10 wk of age, a time equivalent to stage 2 T1D in humans. For tolerance approaches to effectively restore insulin function, patients will require a critical mass of functional β cells. T1D is staged based on presence of β cell–specific Abs and clinical signs of disease: stage 1 T1D patients have two or more unique β cell–specific Abs with normoglycemia and no clinical symptoms; stage 2 T1D patients progress to dysglycemia and no clinical symptoms; and stage 3 T1D patients progress to dysglycemia and symptomatic disease (44). The combination of T cell and B cell activation results in β cell loss and inflammatory lesions within pancreatic islets (insulitis). Studies in NOD mice (28) and humans (29, 30) suggest that 40–85% of the β cell mass must be destroyed by onset of diabetes (stage 3 in humans). Therefore, tolerogenic therapies administered prior to complete destruction of the β cell mass will serve the greatest benefit for patients. The utilization of the present multiprotein tolerance induction platform may also prove to be efficacious in the preservation of β cell transplants. If true, then therapy would not be limited to prediabetic and early onset T1D patients, but it would also useful for T1D patients with late-stage disease.
In addition to the profound protection from T1D development, CNP-T1D treatment inhibited proinflammatory IFN-γ secretion and increased anti-inflammatory IL-10 secretion, as compared with the control unloaded CNP-treated mice (Fig. 3B–J). CNP-T1D treatment significantly inhibited the secretion of IFN-γ upon culture of splenocytes in the presence of the primary epitopes contained within CNP-T1D, for example, insulin, GAD65, and chromogranin A, as well as the response to a nontargeted bystander epitope on IGRP. Additionally, CNP-T1D treatment significantly increased the secreted level of the regulatory cytokine, IL-10, upon culture of splenocytes in all ex vivo activation conditions. These results suggest that CNP-T1D treatment induced a regulatory phenotype within the spleen. Tolerance thus appears to be driven by both intrinsic Ag-specific anergy and resultant prevention of T1D by increasing the numbers of Foxp3+ regulatory CD4+ T cells, IL-10–producing CD4+ T cells (Tr1), and PD-1+ T cells in the spleen as well as PD-L1+ and CD206+ macrophages (Figs. 4C, 4E, 4F, 5E) (14). CNP-T1D treatment significantly decreased the number of both total CD45hi cells, CD4+ T cells, and CD8+ T cells within the pancreas (Fig. 5A, 5D), supporting the hypothesis that although not every potential diabetogenic protein was contained within the CNP-T1D nanoparticle, T cells specific for spread epitopes were not activated following treatment. This hypothesis is supported by the CNP-T1D–induced decrease in the level of IFN-γ secreted upon culture of splenocytes in the presence of IGRP206–214, not contained within CNP-T1D (Fig. 1C). Additionally, of the CD4+ T cells still present within the pancreata of CNP-T1D treatment NOD mice, an increased percentage of these CD4+ T cells was of a regulatory/tolerogenic phenotype (Foxp3+, CTLA-4+, PD-1+, and IL-10+) (Fig. 5G). Likewise, an increased percentage of CD4+ T cells within the pancreatic lymph nodes of CNP-T1D–treated NOD mice was CTLA-4+ (Fig. 5F). These data demonstrate that CNP-T1D induces tolerance to multiple diabetogenic epitopes in the spontaneous NOD model by inducing Treg cell responses. Notably, we have recently published that protection in NOD.scid recipients of activated transgenic BDC2.5 T cells induced by tolerization with PLGA nanoparticle encapsulating the chromogranin p31 mimotope peptide was reversed by both depletion of Treg cells using anti-CD25 (16) and administration of anti–IL-10 (45).
The major advantage of our nanoparticle tolerance delivery system is that induction and maintenance of unresponsiveness is due to the activation of autoantigen-specific Treg cells. Most current Food and Drug Administration–approved therapies for the treatment of autoimmune disease focus on the reduction of disease symptoms using anti-inflammatory drugs, for example, corticosteroids or biologicals (e.g., TNF-α inhibitors) (46). However, long-term use of these drugs is associated with adverse side effects such as increased susceptibility to opportunistic infections and tumors (47). Additional T1D therapies in current development (e.g., anti-CD3) are broadly suppressive and transient and do not treat the root cause of disease by specifically targeting only autoantigen-specific pathogenic T cells. Treg cells are recognized to play a central role in the maintenance of peripheral self-tolerance. Experimental evidence from preclinical animal models has shown that transfer of natural Treg cells can regulate immune responses to self-antigens (48–50). To date, the translation of in vitro–expanded Treg cells for treatment of autoimmune diseases has been hampered by issues such as production efficiency, stability, and the risks of pan-suppression. Ag-specific Treg cells have been shown to preferentially suppress Ag-specific CD4+ T cell responses and to have enhanced disease-associated regulatory function in comparison with polyclonal Treg cells. For example, similar to other findings (51–53), we have shown that 10-fold fewer induced PLP139–151-specific Treg cells (2 × 105) (54) were sufficient to suppress PLP139–151-induced EAE in SJL/J mice as compared with polyclonal natural Treg cells where a minimum of 2 × 106 cells were required (48).
Immune tolerance to circulating Ags and cells is carried out in the spleen and liver (55, 56), as these organs filter the blood and contain phagocytic myeloid cells that evolved to clear and dispose of the large number of apoptotic cells generated by normal cell turnover in the hematopoietic system. Efferocytosis is the process by which dead, dying, and apoptotic cells are cleared from the blood, and this process is vital for the maintenance of immune homeostasis. It has been hypothesized that dysregulation of this process may be involved in several autoimmune disease states (57, 58). We have previously shown that Ag-encapsulated carboxylated PLGA nanoparticles serve as surrogates for apoptotic bodies and following i.v. infusion are specifically taken up by APCs in the splenic marginal zone and liver via the MARCO scavenger receptor (59), which we have shown is necessary for induction of carboxylated PLGA nanoparticle tolerance (15). MARCO has been shown to be responsible for uptake of apoptotic cells (60) and negatively charged nanoparticles (61). Following MARCO engagement, the APCs upregulate PD-L1 and secrete IL-10 and TGF-β, and the individual encapsulated diabetogenic proteins within CNP-T1D are processed and multiple epitopes from each protein are presented to both activated and naive Ag-specific T cells in the tolerogenic environment resulting in anergy induction (23) and the activation of induced Treg cells and Tr1 cells (14). The Treg cells then inhibit the autoimmune T cell responses, restoring overall immunologic homeostasis.
In summary, CNP-T1D treatment effectively and efficiently prevents induction and progression of spontaneous T1D in NOD mice, preventing autoimmune destruction of insulin-producing pancreatic β cells in an Ag-specific manner. Compared to other tolerance-inducing approaches tested in NOD mice (62–65), CNP-T1D–induced tolerance requires fewer treatments and generates more durable protection. Mechanistically, CNP-T1D induces the activation of Treg cells and Tr1 cells that regulate diabetogenic T cell responses to both epitopes contained within the encapsulated β cell proteins as well as bystander suppression, that is, blocking IFN-γ production by diabetogenic T cells specific for nontargeted Ags (IGRP and ZnT8). CNP-T1D appears to be a highly promising approach for translation to testing in human T1D in light of the ease of GMP production of Ag-encapsulating PLGA nanoparticles and the demonstrated success of this Ag delivery platform in inducing gliadin-specific tolerance in a phase 1/2a human celiac disease clinical trial (20).
We thank the Northwestern Mouse Histopathology & Phenotyping Core for their services. We acknowledge the support from all members of the Miller laboratory.
This work was supported by the COUR Pharmaceutical Development Company, Inc. (02-14-2022-4) (grants to S.D.M.), the National Institutes of Health (R01 AI155678), the Juvenile Diabetes Research Foundation (2-SRA-2018-566-S-B), the David and Amy Fulton Foundation, and the Cramer Family Foundation. T.N. was supported by Juvenile Diabetes Research Foundation Postdoctoral Fellowship 3-PDF-2018-582-A-N.
J.R.P., M.-Y.C., T.N., T.M., A.E., and S.D.M. formed the hypotheses, designed experiments, performed experiments, analyzed results, and prepared the manuscript. T.N., S.G., S.K., and M.T.B. helped design experiments, analyze results, and synthesized the CNP-T1D nanoparticles. All authors read and edited the manuscript.
The online version of this article contains supplemental material.
Stephen D. Miller is a Distinguished Fellow of AAI.
Abbreviations used in this article
CNPs encapsulating insulin, chromogranin A, and GAD65
diphtheria toxin A
experimental autoimmune encephalomyelitis
glutamic acid decarboxylase 65
heat shock protein
islet-specific glucose-6-phosphate catalytic subunit-related protein
macrophage receptor with collagenous structure
myelin oligodendrocyte glycoprotein
type 1 diabetes
zinc transporter 8
This study received funding from COUR Pharma, Inc., which is commercializing clinical translation of the Ag-PLGA nanoparticle tolerance technology for the treatment of autoimmune and allergic diseases. S.D.M. is a co-founder, member of the scientific advisory board, holds stock options, and is a consultant for COUR. J.R.P., S.G., S.K., A.E., and M.T.B. are employees of COUR. The other authors have no financial conflicts of interest.