A Microsphere-Based Vaccine Prevents and Reverses New-Onset Autoimmune Diabetes

Three phosphorothioated (*) antisense oligonucleotides targeted to the CD40, CD80, and CD86 primary transcripts were synthesized by Integrated DNA Technologies (Coralville, IA). The antisense oligonucleotide sequences are CD40-antisense, 5′C*A*C* A*G*C* C*G*A* G*G*C* A*A*A G*A*C* A*C*C* A*T*G* C*A*G* G*G*C* A-3′; CD80-antisense, 5′-G*G*G* A*A*A G*C*C* A*G*G* A*A*T* C*T*A* G*A*G* C*C*A* A*T*G* G*A-3′; and CD86-antisense, 5′-T*G*G* G*T*G* C*T*T* C*C*G* T*A*A* G*T*T* C*T*G* G*A*A* C*A*C* G*T*C-3′. Scrambled antisense oligonucleotides were also formulated into microspheres and used as nonsense controls in several experiments.

An aqueous solution of the oligonucleotide mixture was prepared by combining aliquots of three oligonucleotide solutions to form a 10 mg/ml solution. Ten milligrams per milliliter poly-l-lysine-HBr in diH₂O (poly-l-lysine-HBr with an average molecular weight of 50,000 Da; Bachem, King of Prussia, PA) was prepared. Poly-l-lysine-HBr was added to the oligonucleotide solution at a volumetric ratio of 1:1. The mixture was vortexed gently. A 25% polymer solution was prepared containing 12.5% polyvinyl pyrrolidone (PVP; with an average molecular weight of 40,000 Da; Spectrum Chemicals, Gardena, CA) and 12.5% polyethylene glycol (PEG; with an average molecular weight of 3,350 Da; Spectrum Chemicals) in 0.1 mol/l sodium acetate (Spectrum Chemicals) at pH 5.5. The polymer solution was added in a 2:1 volumetric ratio as follows: 750 μl antisense oligonucleotides, 750 μl poly-l-lysine-HBr, 3.0 ml PEG/PVP, to a final total volume of 4.5 ml.

The 4.5-ml preparation was incubated for 30 min at 70°C and then cooled to 23°C. The solution became turbid on cooling, and a precipitate was formed. The suspension was then centrifuged, and the excess PEG/PVP was aspirated. The resulting pellet was washed by resuspending the pellet in deionized water, followed by centrifugation and removal of the supernatant. The washing process was repeated three times. The aqueous suspension was frozen and lyophilized to form a dry powder of microspheres comprising oligonucleotide and poly-l-lysine. Particle size was determined using dynamic light scattering (LB-550 Nanoparticle Size Analyzer; Horiba, Irvine, CA). Microsphere morphology was examined by scanning electron microscopy (S-4800; Hitachi, Pleasanton, CA).

The weight percent load of the antisense oligonucleotide components in the microsphere was determined using gradient reverse-phase high-performance liquid chromatography (HPLC) with UV detection at 260 nm (Waters, Milford, MA). The microspheres were deformulated using competitive displacement of the DNA oligonucleotides from the poly-l-lysine using an excess of poly-l-aspartic acid sodium salt (molecular weight 5,000–15,000 Da; Sigma) in Tris EDTA buffer (pH 7.8) at 55°C for 24 h. The reverse-phase HPLC was performed on a Waters XTerra MS C18 column (4.6 × 50 mm). Mobile phase A was 8.6 mmol/l tetraethylammonium and 100 mmol/l HFIP, pH 8.2. Mobile phase B was methanol. The oligonucleotides were eluted with a 30-min gradient of 15% B to 18% B at a flow rate of 0.5 ml/min. The column temperature was maintained at 60°C.

Approximately 1.1 mg microspheres was suspended into 1.1 ml 1× PBS (pH 7.4) to measure in vitro release. The microspheres were centrifuged at several time points, and the release medium was aspirated and measured at UV 260 nm. Fresh release medium was added to the microspheres until the next time point. The release studies were conducted at 22 and 37°C.

Female NOD/LtJ, NOD-scid, C57BL/6, and Balb/c mice were purchased from The Jackson Laboratories (Bar Harbor, ME) and used between the ages of 5 and 22 weeks. Animals were maintained in a specific pathogen-free environment in the Animal Facility of the Rangos Research Center and used in full compliance with experimentation protocols approved by the Animal Research Care Committee of the Children 's Hospital of Pittsburgh.

All biochemical and cell culture reagents were purchased from Invitrogen (Carlsbad, CA). Antibodies were purchased from BD Biosciences (San Diego, CA) either as directly conjugated fluorescent embodiments or as affinity-purified preparations in concert with fluorescently labeled isotype-matched secondary products. The specific clones used were CD4 clone, RM4-5; and CD25 clone, 7D4. To ascertain the prevalence of Foxp3⁺ cells, we used the kit commercially available from eBioscience (San Diego, CA), which includes the FJK-16S Foxp3 clone. Spleen or lymph node–derived cells were enriched into T-cells using column methodology (R&D Systems, Indianapolis, IN). The NIT-1 (NOD insulinoma T-antigen–transformed-1 cell line) insulinoma cell line (CRL-2055; American Type Culture Collection) was maintained in Ham's F-12K medium with 2 mmol/l l-glutamine and 10% heat-inactivated fetal bovine serum until 70% confluence. The cells were then gently removed by collagenase digestion and made into lysates by repeated freeze-thaw cycles. The lysate was dispensed into aliquots in sterile PBS.

Microspheres formulated with the mixture of the CD40, CD80, and CD86 antisense oligonucleotides (AS-MSPs) or with scrambled control sequences (SCR-MSP) were dispensed into aliquots of 50-μg oligonucleotide formulations in sterile PBS. One hundred microliters of AS-MSP, SCR-MSP, or PBS was injected subcutaneously, at a site anatomically proximal to the pancreatic lymph nodes, into 5- to 8-week-old NOD female mice in the initial prevention study. To determine their efficacy in new-onset diabetic NOD mice, we first treated diabetic mice (determined by two consecutive blood glucose measurements of >300 mg/dl) once daily with 2 units 1:1 Humulin R:Humulin N mix in PBS until nonfasting blood glucose stabilized to below 300 mg/dl. Insulin treatment was immediately stopped, and the microsphere formulations were injected subcutaneously, at a site anatomically proximal to the pancreatic lymph nodes, three times a week until the mice were killed for further study.

To determine whether the microsphere administration induced regulatory immune cell populations, we isolated spleen from diabetes-free NOD mice, administered the AS-MSP, and made single cells. Parallel single-cell preparations were enriched into T-cells. Splenocytes and enriched T-cells were cotransferred in equal numbers (1 × 10⁷) into 7- to 10-week-old female NOD-scid recipients by intravenous injection. Diabetes was monitored on consecutive days, twice weekly.

Fluorescence-activated cell sorter (FACS) analyses were performed on spleen or lymph node cells of successfully treated NOD mice periodically. These were mice that were treated at <10 weeks of age and remained diabetes free or new-onset diabetic mice that were “reversed.” Specifically, cell surface phenotype of the T-cells was assessed by FACS analysis in a FACSVantage SE instrument using FACSDiva and CellQuest modules (BD Biosciences). In addition to the antibodies described above, relevant isotype-specific, fluorescently conjugated antibodies were used throughout as controls for nonspecific cell surface binding. Cells were incubated with mixtures of specific antibodies, and in parallel, with the isotype controls at titers between 1:100 and 1:500 as per the manufacturer's suggestions. They were also labeled with propidium iodide, 7-AAD, and/or the annexin V-staining reagent (Invitrogen-Molecular Probes) and then used directly for FACS analysis. Percentage of positive cells or mean fluorescence intensity was measured for cells gated on forward and side- scatter properties representing T-lymphocytes. Dead cells and cell clumps were excluded from the analyses, which were performed using the CellQuest software package.

To ascertain proliferation of T-cells from AS-MSP–treated NOD mice to alloantigens or nominal antigens, spleens were isolated from randomly selected mice and enriched into T-cells. These were then cocultured with equal numbers of irradiated splenocytes (1 × 10⁵) from allogeneic mice or syngeneic mice in the presence/absence of 1 μg intact ovalbumin. Proliferation was measured 5 days later using the CyQuant fluorometric reagent (Invitrogen-Molecular Probes) as directed by the manufacturer. The coculture supernatants were retained to measure cytokine levels by Luminex-based fluorescence methods (Beadlyte; Upstate Biotechnology). To measure proliferation in response to the NIT-1 cell line-derived lysate, 1 × 10⁵ splenocytes from treated NOD mice were cocultured with an equal number of syngeneic and age-matched irradiated splenocytes with or without the addition of 5 μg NIT-1 cell lysate. After 5 days, the supernatant was removed for cytokine analysis, and proliferation was measured.

The degree of insulitis in the pancreata of randomly selected successfully treated NOD mice was ascertained in serial sections of formalin-fixed tissue by hematoxylin-eosin treatment. Additional sections were probed for insulin content using a commercially available method (Vector BioLabs) with an insulin antibody (DakoCytomation, Carpinteria, CA).

NOD female mice (8 weeks of age) were anesthetized with isoflurane delivered by the XGI-8 Gas Anesthesia System (Xenogen). Initial isoflurane concentration was set to 2.5% and was reduced to 1.5% once the animal was anesthetized. Mice were imaged before injection on the IVIS Lumina (Xenogen) system using a dsRed filter set. Fluorescent exposure times of 0.8 s were used for in vivo imaging and of 0.1 s for ex vivo imaging. Once imaged for background fluorescence, mice received a 300-μl intraperitoneal injection that contained either a mixture of 150 μl fluorescent microspheres (FluoSpheres 580 nm/605 nm; Molecular Probes-Invitrogen) and 100 μl PBS (control) or a mixture of 150 μl fluorescent microspheres (FluoSpheres 580 nm/605 nm) and 0.5 μg/μl AS-MSP microspheres. Mice were then imaged just before euthanasia (and organ excision) at 3, 24, and 48 h after injection. The spleen and pancreas were then removed and imaged immediately afterward. In a subsequent study, a PROMAXX formulation of a Cy3-labeled oligonucleotide nontargeting siRNA sequence purchased from Dharmacon (Lafayette, CO) was injected subcutaneously at an anatomically distal and proximal site to the pancreatic lymph nodes (scruff and flank, respectively) and the pancreas and spleen were excised 3 h after the injection. Average radiance of organs was calculated using Living Image version 3.0 (Xenogen-Caliper Life Sciences), and mean values were graphed.

Female NOD mice (8–12 weeks of age) were treated with fluorescent microspheres 0.2 μm in diameter with the excitation emission profiles of 505 nm/515 nm (FluoSpheres 505 nm/515 nm; Molecular Probes-Invitrogen). The mice were injected subcutaneously with a mix of either 100 μl fluorescent microspheres and 100 μl PBS (control) or 100 μl fluorescent microspheres and 100 μl 0.5 μg/μl antisense oligonucleotide mixture (CD40, CD80, and CD86) in a total volume of 300 μl. Spleens where excised at 1, 2, 3, and 6 days after injection. Single splenocytes were treated with Mouse BD Fc block (BD Pharmingen) for 5 min to reduce nonspecific antibody binding. Cells were then treated with anti-mouse CD11c allophycocyanin (APC) and CD40 phycoerythrin (PE), CD80 PE, or CD86 PE for 30 min (BD Pharmingen), after which the cells were washed and fixed in a 2% paraformaldehyde solution. Cells were then gated by FACS into CD11c and fluorescent microsphere double-positive cells, and within this gate, CD40, CD80, and CD86 levels were further ascertained. Control isotypes were used throughout this experiment.

Student's t test, ANOVA, and Kaplan-Meier log-rank analysis where described were facilitated using the Prism version 4 software by GraphPad (San Diego, CA).

Scanning electron micrographs of the PROMAXX microspheres exhibited a relatively smooth surface (Fig. 1A). The particle size of the microspheres determined by light scattering was 0.5–4 μm in size with an average particle size of ∼2.5 μm. Thus, the calculated surface area of a single microsphere is ∼19.6 μm². The loading of oligonucleotides determined by reverse-phase HPLC in the microspheres was ∼70% weight by weight. Based on these measurements, the calculated number of copies of antisense oligonucleotide per microsphere was ∼1.05 × 10⁸.

The in vitro release kinetics of the antisense oligonucleotides from the microspheres is shown in Fig. 1B at 22 and 37°C. The data show that after a small initial burst effect, the cumulative percent release is proportional to the square root of time. However, after 120 h, <1.1% of the incorporated antisense oligonucleotides have been released at 37°C and <0.8% release was observed at 22°C. This suggests that despite the 70% weight percent loading of oligonucleotide in the microspheres, most of the drug release occurs only after the microsphere has been uptaken by cells in vivo.

Our previous studies demonstrated that NOD-derived dendritic cells treated ex vivo with a mixture of phosphorothioated antisense oligonucleotides targeting the 5′ end of the CD40, CD80, and CD86 primary transcripts prevented diabetes in syngeneic recipients (19,20). To determine whether the microsphere-formulated antisense mixture was as efficacious, we administered AS-MSP targeting the 5′ end of the CD40, CD80, and CD86 primary transcripts into NOD female mice between the ages of 5 and 8 weeks. As controls, we concurrently treated parallel groups with PROMAXX formulation containing SCR-MSP and PBS vehicle. One single injection of AS-MSP, at a site anatomically proximal to the pancreatic lymph nodes, significantly delayed onset of diabetes (Fig. 1C), and eight consecutive injections (Fig. 1D) were very efficacious in preventing the disease altogether. All NOD mice treated with control formulations and untreated NOD mice developed diabetes by 22 weeks of age. We are now determining whether fewer consecutive injections can be as efficacious.

We then proceeded to determine whether the AS-MSP could reverse new-onset hyperglycemia, which could suggest reversal of autoimmunity and mechanisms promoting preservation of residual β-cell mass in NOD mice. NOD mice between 12 and 16 weeks of age developed diabetes confirmed by two consecutive blood glucose readings of >300 mg/dl. These mice were treated with daily insulin injections intraperitoneally until glycemia was stabilized to <300 mg/dl. Decreased blood glucose was generally observed over a 10- to 18-day period of insulin administration. The insulin was then immediately discontinued and AS-MSP, SCR-MSP, or PBS was injected subcutaneously, at a site anatomically proximal to the pancreatic lymph nodes, twice weekly for no more than 25 days after the first microsphere administration (Fig. 2A). Figure 2B and C demonstrate that AS-MSP administration can reverse new-onset hyperglycemia that is stably maintained even after cessation of the microsphere treatment. This outcome is reproducible, and in Fig. 2D, we show the outcome from three additional study groups. An additional 7 of 15 mice exhibited reversal of hyperglycemia and stable maintenance after cessation of the AS-MSP administration.

To ascertain the degree to which pancreatic inflammation was affected in NOD recipients of the AS-MSP that exhibited long-term protection from diabetes (treated with AS-MSP at <8 weeks of age; cohort shown in Fig. 1D), we isolated pancreata from three randomly selected AS-MSP recipients, SCR-MSP controls, and diabetic NOD mice. Whereas SCR-MSP–treated mice and diabetic controls exhibited significant insulitis and indistinguishable islet mass, respectively, we observed normal islet architecture with an absence of insulitis in AS-MSP recipients (Fig. 3).

Our previous data suggested that the ex vivo antisense-treated dendritic cells were suppressive at least in part by augmenting the prevalence of CD4⁺ CD25⁺ putative regulatory T-cells (Treg cells) (19,20). We have now specifically addressed the question of whether AS-MSP directly augment Treg cells by enumerating the percentage of Foxp3⁺ CD25⁺ T-cells inside a CD4⁺ gate in the spleen and lymph nodes of pre-diabetic NOD females. AS-MSP alone, and not the control microspheres or PBS, augmented the prevalence of Foxp3⁺ CD25⁺ T-cells (Fig. 4).

Given that the AS-MSP treatment yielded augmented numbers of CD4⁺ CD25⁺ Foxp3⁺ putative Treg cells, it was logical to posit that a splenocyte population from AS-MSP–treated mice should be able to confer some degree of suppression to diabetes inducement in immunodeficient mice administered splenocytes from diabetic NOD mice. Toward this objective, a 1:1 mix of splenocytes from AS-MSP–treated diabetes-free NOD mice and new-onset diabetic NOD mice was injected into female NOD-scid mice between 6 and 10 weeks of age. As controls, a 1:1 mix of splenocytes from nondiabetic 10-week-old NOD mice was treated with either SCR-MSP (eight consecutive subcutaneous injections spaced every 3 days apart), and splenocytes from diabetic mice were injected into 6- to 10-week-old NOD-scid females. In addition, another control group of NOD-scid mice was injected with splenocytes from new-onset diabetic NOD mice. Each cell population consisted of 1 × 10⁷ splenocytes freshly isolated. Blood glucose was monitored twice weekly, and two consecutive readings of nonfasting glucose >300 mg/dl was considered as the diabetic threshold. Figure 5 demonstrates that adoptive transfer of diabetes in NOD-scid recipients was almost completely abrogated in the presence of splenocytes derived from AS-MSP recipients that were diabetes free.

In vivo–injected AS-MSPs accumulate within the pancreatic lymph nodes and spleen and confer decreased costimulatory molecule surface expression on splenic dendritic cells in vivo. To determine the route of migration of the AS-MSP, which could offer insight into the mechanism of immunoregulation and diabetes suppression, we used in vivo imaging technology to observe the temporal accumulation of a 1:1 mix of commercially available fluorescent microspheres and the AS-MSPs. The sizes of the commercially purchased fluorescent microspheres (FluoSpheres) were considerably smaller than the particle size of the AS-MSP. The AS-MSP were 2.5 μm on average and the FluoSpheres were 0.2 μm on average. Thus the FluoSpheres should not interfere with the uptake of the AS-MSP by dendritic cells. In Fig. 6, we show that as early as 3 h after intraperitoneal injection, the fluorescent microspheres accumulated at anatomical sites where the pancreas and the spleen reside (Fig. 6A). The intensity of the fluorescence did not change over a 72-h monitoring period (data not shown). To confirm the accumulation of the microspheres within the pancreas and spleen, we excised these two organs from the mice that were being monitored live, and in Fig. 6B, we confirm that the fluorescent microspheres accumulated as early as 3 h after injection at distinct foci in pancreas, which we interpret to be the pancreatic lymph nodes, and at distinct foci within the spleen. Although accumulation of fluorescent microspheres was observed in these two organs regardless of whether AS-MSP were included in the mix or not, the actual intensity of the fluorescence was different in organs excised at the times indicated in Fig. 6C and D. The fluorescence was evident in the spleen and pancreata of all mice at all time points studied (from 3 h to 6 days after injection) (Fig. 6; data not shown).

As an additional confirmation that oligonucleotide-formulated microspheres accumulate inside the pancreas and spleen, we injected PROMAXX-formulated Cy3-labeled siRNA to CD86 identically as described above. We also compared the pancreatic accumulation of this Cy3-labeled formulation after intraperitoneal injection, subcutaneous injection at a site distal to the anatomic location of the pancreatic lymph nodes (at the scruff of the mouse), and subcutaneous injection at the site used in the prevention and reversal studies documented herein (flank of the mouse). Figure 6E confirms the pancreatic accumulation of PROMAXX-formulated Cy3-labeled oligonucleotide at 3 h after injection but only when the subcutaneous injection was performed anatomically proximal to the pancreatic lymph nodes.

To confirm that AS-MSP administration conferred a decrease in CD40, CD80, and CD86 in dendritic cells in vivo, we injected NOD mice intraperitoneally with a mixture of the FluoSpheres with PBS as control or with AS-MSP as described in research design and methods. At the times indicated at the top of each graph in Fig. 6F, we excised the spleen and measured CD40, CD80, and CD86 levels in CD11c⁺ fluorescence⁺ double-positive single cells. We observed a decreased level of all three costimulatory molecules on CD11c⁺ cells that had concentrated the mixture of fluorescent microspheres and AS-MSP as early as 1 day after injection with maintenance of these levels compared with CD11c⁺ cells from control mix-treated NOD mice over a 6-day monitoring period. The only exception was seen at day 3 for CD40.

A major concern for eventual translation of diabetes-suppressive therapies into human trials is the antigen specificity and therefore the cell specificity of the approach and whether the treatment confers global and nonspecific suppression. To address these issues, we killed randomly selected diabetes-free mice from the cohorts shown in Fig. 1D, and we then proceeded to ascertain the proliferation of splenic and lymph node T-cells to alloantigens, nominal antigens using intact ovalbumin and also using syngeneic β-cell–derived antigen in the form of cell lysate from the NOD-derived insulinoma cell line NIT-1 (24,25). Although insulin and GAD are viable candidate autoantigens with mechanistic and teleological involvement (26), the nature of the initiating autoantigen remains unclear. Nevertheless, it is reasonable to consider that it should be β-cell resident. Therefore, we used the NIT-1 cell line that derives from an NOD insulinoma as a source of β-cell antigen in cocultures of T-cells from diabetes-free NOD mice treated with the AS-MSP to determine the possibility of antigen-specific hyporesponsiveness. Supplementary Fig. 7 (available in an online appendix at http://dx.doi.org/10.2337/db07-0507), shows that T-cell proliferation to nominal and alloantigen is maintained, whereas there is T-cell hypoproliferation in cocultures with NIT-1 cell lysate. Furthermore, ascertaining the cytokine profile in the coculture supernatants, we observed a significant decrease in TNF-α production by T-cells from AS-MSP–treated, diabetes-free NOD mice even in the presence of NIT-1 lysate (Supplementary Fig. 7F; from those diabetes-free NOD mice in Fig. 1D). Although IFN-γ production was slightly decreased in the cocultures of T-cells from the AS-MSP–treated mice, it was not statistically distinguishable from the cocultures with T-cells from PBS-treated mice in the presence of NIT-1 lysate. The assay, finally, could not detect the presence of IL-4, IL-10, or TGF-α in the supernatants.

Formulation of bioactive agents into microparticles offers a versatile means of delivering these molecules in vivo especially for the purpose of modulating the immune system (21,27). An important component of the modulatory properties of microspheres is the polymer backbone that often stimulates potent antigen-presenting cell activation, which is beneficial for tumor immunotherapy or antipathogen interventions (28,29). For the purposes of immunosuppression, however, the polymer and the component chemistries should be such that at the very minimum, they should be neutral on antigen-presenting cell state. PROMAXX nucleic acid microspheres offers this versatility (23) as a result of the minimal quantity of formulation excipients. In this regard, we now show its utility as a key component of a diabetes-suppressive vaccine.

The most noteworthy finding in these studies is the capacity of the AS-MSP to reverse new-onset hyperglycemia, which we believe is underlined by preservation of residual β-cell mass. Nevertheless, we cannot yet formally distinguish this from a potential regenerative process of β-cells (division of existing β-cells or differentiation of progenitor/stem cells from ductal epithelium [30–32]). There is, however, considerable support from previous studies for a sufficiency in β-cell mass that is functionally responsive to glucose at the time of diagnosis of type 1 diabetes (32–36). This mass, if permitted to recover from autoimmune attack by modulating the aggressive and β-cell–specific immune cells, may provide the patient with normoglycemic metabolic control.

Although our data herein are not the first to demonstrate small molecule–based diabetes prevention in the NOD mouse, they are the first to show a well-defined microparticle system that can reverse hyperglycemia in new-onset disease whose mechanism of action is decipherable. In preliminary studies, we have observed that when the formulation was injected into NOD-scid immunodeficient mice reconstituted with splenocytes from NOD mice of various diabetes stages (young, 12–15 weeks old where autoimmunity is already established and where β-cell function is impaired, and from new-onset diabetics), the prevalence of CD4⁺ CD25⁺ putative Treg cells increased. We have now confirmed herein in a more direct experiment that CD4⁺ CD25⁺ Foxp3⁺ T-cells numbers increase in female NOD mice administered AS-MSP. These findings may explain the capacity of T-cells from AS-MSP–treated NOD mice free of diabetes to prevent the adoptive cotransfer of the disease to NOD-scid recipients. More importantly, T-cells from diabetes-free AS-MSP exhibited poor proliferation to NIT-1 lysate in vitro while proliferating vigorously in cocultures with allogeneic irradiated splenocytes or when pulsed with ovalbumin.

Many lines of evidence conclude that injected microsphere formulations are rapidly taken up by resident and/or migratory antigen-presenting cells, especially dendritic cells, and accumulate inside lymphoid organs anatomically proximal to the site of injection. Our studies in the past have also confirmed this uptake/trafficking (K.N., J.H., N.G., unpublished observations). Herein, we have provided some of these data, which confirm that subcutaneously injected microspheres accumulate as early as 3 h after injection within the pancreas, very possibly in the lymph nodes, and eventually in the spleen. Although we do not explicitly demonstrate it, we are very confident that the most likely method of microsphere accumulation inside the pancreas is via dendritic cells. Of wide interest is the mechanism by which antigen and therefore cell/tissue specificity is acquired by the microsphere-loaded dendritic cells. A series of elegant studies (12–14,17) point to a process in which migratory dendritic cells potentially acquire the microspheres that are injected physically proximal to a site of ongoing inflammation and, guided by pro-inflammatory signals deriving from the diabetic pancreas, acquire pancreatic β-cell antigens in the form of apoptotic cells at a site of inflammation. Then, these dendritic cells will exit the inflamed tissue and accumulate within the regional lymphoid organs where they can engage not only effector T-cells but Treg cells as well (12–14,17,33). It has been shown that dendritic cells that acquire apoptotic cells enter into a state of functional immaturity, which may result in tolerance to the acquired antigens (17,34). We hypothesize that a similar, if not identical, process is occurring immediately after AS-MSP administration in NOD mice. The inflammation inside the pancreas with associated β-cell apoptosis will drive AS-MSP–loaded dendritic cells to acquire β-cell antigen, and immediately thereafter, their accumulation inside the pancreatic lymph nodes will facilitate their interaction with Treg cells, which may, in themselves or in concert with AS-MSP–stabilized dendritic cell or other endogenous dendritic cell subsets, induce β-cell–specific immune hyporesponsiveness or functional tolerance to β-cell–restricted antigens (35–37). We show that microsphere-loaded dendritic cells in the spleen express decreased levels of CD40, CD80, and CD86 at their surface, and we anticipate that this is true for dendritic cells in the pancreatic lymph nodes. Although we do not currently have an explanation for the increase in CD86 levels in fluorescent dendritic cells from the spleen of mice treated with the AS-MSP mixture at day 6 after administration, this could reflect the eventual degradation of the antisense nucleic acid in the dendritic cells by day 6, the disappearance of the migratory dendritic cells and acquisition of the fluorescent particles by endogenous, secondary dendritic cells in a cross-priming mechanism, or both.

What has become apparent in these preliminary accumulation studies is the differential accumulation of microspheres inside the pancreas after subcutaneous injection depending on the proximity of the injection site to the pancreas. When administered subcutaneously at a site that flanks the site of the pancreatic lymph nodes, the microspheres accumulate within 3 h. In contrast, there is no detectable accumulation when the microspheres are administered subcutaneously at a site that does not drain to the pancreatic lymphatics. This observation is relevant for two reasons. First, it offers insight into which administration route may be more clinically useful. Second, it suggests that immunoregulatory interventions, at least those that involve dendritic cells as intermediates, are mechanistically active only when the pancreatic lymph nodes are involved (i.e., the nexus of autoimmune cells and regulatory cells that exhibit antigen specificity). We are, consequently, very interested in identifying the precise mechanism involved in costimulatory protein surface density changes at the dendritic cells in spleen and in pancreatic lymph nodes in response to the AS-MSP treatment. The scarcity of dendritic cells from the pancreatic lymph nodes at the times shown in Fig. 6 did not permit us to pursue an effective enrichment of these cells, but it is one of our objectives in future studies in which the precise mechanisms and temporal sequence of cell migration and interaction with other cells in the pancreatic lymph nodes and/or at the islets of Langerhans remain to be established.

Many investigators support immunotherapy approaches for autoimmunity where putative autoantigen supply provides the antigen, and hence the tissue, specificity (38–41). It is worth noting, however, that suppression of autoimmune disease in animal models need not require antigen supply (37). Mechanistically, such interventions may involve bystander tolerance, linked suppression, or similar phenomena (37,42,43). Although we did not supply autoantigen to the AS-MSP regimen, we did inject diabetic NOD mice with insulin to normalize the glycemia. In one possible mechanism, the exogenous insulin supply (as a well-characterized putative autoantigen) before AS-MSP administration may be acting similar to or in an identical manner as earlier insulin-based tolerance strategies (1,38–41) . However, we think a more likely mechanism is that the AS-MSPs stabilized subsets of endogenous dendritic cells toward diabetes-suppressive states by downregulating cell surface CD40, CD80, and CD86. We are currently actively investigating the possible mechanism experimentally. It must be stressed, concurrently, that insulin administration alone cannot account for the AS-MSP effects in its physiological glucoregulatory capacity. No mouse in groups treated with insulin alone was capable of maintaining normoglycemia for more than 1 week after withdrawal of the insulin (Fig. 5). This argues against the possibility that in the AS-MSP–treated animals, new-onset diabetes reversal was due to insulin-induced β-cell rest phenomena (44,45). Taken together, these findings may be readily translatable clinically with an immediate aim of preserving residual β-cell mass in newly onset or preclinical human autoimmune diabetes.

This work was supported in part by the National Institutes of Health (grant DK063499 to M.T. and N.G.), the Juvenile Diabetes Research Foundation (grant 17-2007-1066 to N.G.), and a sponsored research agreement from Epic Therapeutics, a wholly owned subsidiary of Baxter Health Care Corporation.

William Fowle at the Electronic Materials Research Institute at Northeastern University (Boston, MA) conducted the scanning electron microscopy studies.