Dendritic Cell-Directed CTLA-4 Engagement during Pancreatic β Cell Antigen Presentation Delays Type 1 Diabetes Free

Type 1 diabetes (T1D) is an autoimmune disorder resulting from the specific destruction of insulin-producing pancreatic β cells and subsequent hyperglycemia. T cells, specific for β cell Ags, play a major role in the disease process, and they arise and expand likely because of defective immune regulation (1). Immune regulatory defects and disease susceptibility in humans and in experimental animals have been linked to the genetic loci encoding for CTLA-4 and MHC proteins (2–4).

CTLA-4 is a critical negative regulator of T cell activation. This receptor plays a primary role in T cell homeostasis and peripheral tolerance through the inhibition of T cell activation, IL-2 production, and cell cycle progression as evidenced by the lethal lymphoproliferation seen in CTLA-4–deficient mice (5). The mechanisms by which CTLA-4 can mediate negative regulation of T cell responses include its ability to compete with CD28 for ligand binding, negatively regulate TCR signaling, induce release of the tolerogenic enzyme indoleamine 2,3 dioxygenase in APCs, interfere with the lipid raft formation on the T cell surface, and disrupt CD28 localization at the immunological synapse (6–10). These properties suggest the importance of CTLA-4 in maintaining peripheral tolerance and preventing autoimmunity (11–16). These studies together with recently identified abnormalities in the expression levels and signal strength of CTLA-4 in various autoimmune diseases including T1D (17, 18) have raised the specter of modulating CTLA-4 signaling to treat autoimmune diseases.

Splice variants of CTLA-4 have been identified as potential risk factors contributing to the development of T1D in both patients and NOD mice (2, 4, 17). Although the disease susceptibility is mapped to an allelic variation in the CTLA-4 gene and correlated with mRNA levels of soluble CTLA-4 in humans, increased risk for developing the disease has been correlated with differential mRNA levels of a ligand-independent form of CTLA-4 in NOD mice (2, 17). These reports suggest that inadequate levels of signaling through CTLA-4 could be responsible for inducing and/or promoting T1D, and, therefore, enhancing CTLA-4 signaling strength on self-Ag–presenting APCs could prove effective in preventing autoimmunity.

Recently, using mice that were immunized with specific foreign- and self-Ags, we demonstrated that enhanced engagement of CTLA-4 from Ag-presenting dendritic cells (DCs) induces adaptive T regulatory cells (Tregs) that express Foxp3, IL-10, and TGF-β1 (16). In a different study, we have shown that preferential ligation of CTLA-4 by CD80 upon pancreatic β cell Ag presentation results in the generation of TGF-β1– and IL-10–producing adaptive Tregs (19). These observations along with the noted abnormalities of CTLA-4 signaling in T1D prompted us to examine whether enhancing the strength of CTLA-4 signaling during pancreatic β cell Ag presentation using a DC-directed CTLA-4 engagement method could modulate the autoimmune response in T1D. Because autoimmune diseases occur spontaneously in humans, examining the therapeutic potential of this DC-directed CTLA-4 engagement approach in a spontaneous autoimmune model is important to realize its clinical applicability. Therefore, we have tested β cell Ag-pulsed DCs coated with an agonistic anti–CTLA-4 Ab for their ability to suppress autoreactive T cells and T1D in NOD mice. Our results show that this DC-directed CTLA-4 engagement approach has the potential to suppress insulitis and produce significantly delayed hyperglycemia in NOD mice. Further, this disease-suppressive effect was associated with an increase in the frequencies of Foxp3⁺ and IL-10– and TGF-β1–producing hypoproliferative T cells.

Wild-type NOD/LtJ and NOD.BDC2.5 TCR-transgenic (Tg) mice were originally purchased from The Jackson Laboratory (Bar Harbor, ME), and breeding colonies were established and maintained in the pathogen-free facility of the biological resources laboratory of the University of Illinois at Chicago (Chicago, IL). Glucose levels in the tail vein blood samples of wild-type mice were monitored with the Ascensia Microfill blood glucose test strips and an Ascensia Contour blood glucose meter (Bayer, Pittsburgh, PA). The animal studies were approved by the animal care and use committee of the University of Illinois at Chicago.

Immunodominant β cell Ag peptides [namely: 1) insulin B_(9–23); 2) GAD65_(206–220); 3) GAD65_(524–543); 4) IA-2β_(755–777); and 5) IGRP_(123–145)] and BDC2.5 TCR-reactive peptide (YVRPLWVRME; referred to as BDC peptide) were custom synthesized (GenScript, Piscataway, NJ) as described in the earlier studies (20–26) and used in this study. Peptides 1–5 were pooled at an equal molar ratio and used as β cell Ags for pulsing DCs for in vitro and in vivo experiments as described in our earlier studies (19, 27), unless indicated otherwise.

Hamster anti-mouse CTLA-4 hybridoma (UCI0-4-F-I0-11) and anti-mouse CD11c hybridoma (N418) were purchased from American Type Culture Collection (Manassas, VA) and grown in serum-free/protein-free medium (BD Biosciences, San Jose, CA), and the Abs were purified using Protein L or A (Sigma-Aldrich, St. Louis, MO) columns. Anti–CTLA-4–anti-CD11c bispecific Ab (test BiAb) was prepared by N-succinimidyl-3-(2 pyridyldithiol) propionate (SPDP)-succinimidyl-4-(p-maleimidophenyl) butyrate (SMPB) cross-linking approach as described in our earlier studies (11, 12, 16). Briefly, equal amounts of anti-CD11c and anti–CTLA-4 Abs (in borate-buffered saline [pH 8.5]) were activated using SPDP and SMPB, respectively. The SPDP-activated anti-CD11c Ab was treated with DTT, desalted using a PD-10 column (Sigma-Aldrich), concentrated by ultrafiltration, mixed with SMPB-treated anti–CTLA-4 Ab, and incubated for 4 h. Free active groups of this Ab mixture were blocked with iodoacetamide and purified by gel-filtration chromatography using a Sephacryl S200 column (GE Healthcare, Piscataway, NJ). Purified hamster IgG (Fitzgerald Industries International, Acton, MA) linked to the anti-CD11c Ab similarly served as a control BiAb. Ag-binding efficiencies of BiAbs were tested by FACS using bone marrow-derived DCs (BMDCs) and ELISA using rCTLA-4-Ig (R&D Systems, Minneapolis, MN) as described earlier (16).

Purified anti-mouse-CD16/CD32 (FC block) Abs; FITC-conjugated anti-mouse CD11c, CD4, CD25, IFN-γ, IL-17, TNF-α, and IL-10 Abs; PE-labeled anti-mouse CD80, CD86, CD40, I-Ag⁷, CD4, CD25, CTLA-4, CD28, CD69, TGF-β1, IL-10, and Foxp3 Abs and streptavidin; biotin-labeled and anti-mouse/human TGF-β1 (clone A-75-3), affinity-purified anti-LAP Ab; PE-Cy5–labeled anti-mouse CD4 and CD62L Abs and streptavidin; and PE-Texas Red–labeled anti-mouse CD4 Ab were purchased from Invitrogen (Carlsbad, CA), BD Pharmingen (San Diego, CA), eBioscience (San Diego, CA), R&D Systems, or Biolegend (San Diego, CA), respectively, and used in various studies requiring FACS analyses. Magnetic bead-based cell isolation kits were purchased from Miltenyi Biotec (Auburn, CA). Paired Abs and required cytokine standards for detecting mouse IL-2, IL-4, IL-17, IFN-γ, and IL-10 (eBioscience) and activated TGF-β1 (R&D Systems or BD Pharmingen) were used in ELISA. Multiplex cytokine assay reagents were purchased from BioSource International (Camarillo, CA). The lowest detection limits of Ab pairs and reagents for these cytokines were <10 pg/ml. Whereas either multiplex assay or ELISA method was employed for quantifying most cytokines, activated TGF-β1 was quantified by ELISA method.

DCs were generated in vitro from BM cells and coated with anti–CTLA-4–anti-CD11c or hamster IgG–anti-CD11c BiAb as described in our previous study (16). Briefly, prior to use in some experiments, DCs (1 × 10⁶/ml) were incubated for 48 h at 37°C in the presence of an equal-molar mixture of immunodominant peptides (β cell Ag; 5 μg/ml) or BDC peptide (1μg/ml) and bacterial LPS (5 μg/ml). Cells were washed, incubated with control or test BiAb (10 μg/10⁷ cells/ml) for 30 min on ice, washed further, and used as control-Ab– or anti–CTLA-4–coated DCs for in vitro and/or in vivo experiments. Anti–CTLA-4 Ab- and control-Ab–coated DCs are referred to as anti–CTLA-4-Ab DCs and control-Ab DCs, respectively. β cell Ag-pulsed anti–CTLA-4 Ab-coated DCs and β cell Ag-pulsed control Ab-coated DCs are referred to as β cell Ag-pulsed anti–CTLA-4-Ab DCs and Ag-pulsed control-Ab DCs, respectively. Ab-coated DCs were tested for bound Ab levels by FACS prior to every experiment poststaining with FITC-labeled anti-hamster IgG Ab. Maturation status of DCs was also confirmed using fluorochrome-labeled Abs against activation markers CD80, CD86, CD40, and MHC class II by FACS as described before (16). The Ab-coated DCs were cultured for an additional 36 h and tested for the expression levels of activation markers on the surface by FACS and spontaneously secreted cytokine levels in the culture supernatant by ELISA.

Ag-pulsed Ab-coated DCs (5 × 10⁴ cells/well) were cultured in 96-well round-bottom plates in triplicate along with CFSE-labeled or unlabeled purified CD4⁺ T cells (1 × 10⁵/well) from 8-wk-old NOD.BDC2.5 TCR-Tg mice. After 5 d of culture, CFSE-labeled cells were stained using PE-linked CD4-specific Abs and examined for CFSE dilution by FACS. For some assays, rIL-2 or TGF-β1 or neutralizing Abs against IL-10 and/or TGF-β1 were added to the culture wells. Purified total T cells from diabetic NOD mice were also incubated with β cell Ag-pulsed Ab-coated DCs. After 48 h, cells were pulsed with l μCi/well [³H]thymidine for 18 h. Thymidine incorporation was measured as described before (12). T cell proliferation against ex vivo antigenic challenge was tested by CFSE-dilution assay using splenic and pancreatic LN (PnLN) cells from Ab-coated DC-treated mice as described earlier (19, 26). CFSE dilution was measured by FACS analysis after 5 d of culture.

Female NOD mice of different age groups were injected i.v. with 5 × 10⁶ Ag-pulsed or nonpulsed control or anti–CTLA-4-Ab DCs twice at a 15-d interval and examined for blood glucose levels every week. Two and/or 6 wk postinjection, mice were sacrificed to test for T cell response to ex vivo challenge with the Ag, T cell phenotype, and insulitis. In some experiments, 6-wk-old NOD mice were adoptively transferred (i.v.) with CFSE-labeled Ab-coated DCs (5 × 10⁶ cells/mouse). These mice were euthanized 36 h postinjection, and single-cell suspensions from spleen, PnLN, and pancreas were examined for CFSE⁺ DCs by FACS poststaining with PE-labeled anti-CD11c Ab.

Cells were washed (with PBS supplemented with 2% FBS [pH 7.4]), and Fc receptors were blocked using anti-CD16/CD32 Ab or 5% rat serum on ice for 15 min. For surface staining, cells were incubated with FITC-, PE-, and PECy5- or PE-Texas Red–labeled appropriate Abs in different combinations on ice for 30 min. For intracellular staining, surface-stained cells were fixed, permeablized using a fixation/permeablization kit (eBioscience), blocked using 5% normal rat serum, and incubated with fluorochrome-labeled isotype control or marker-specific Abs at room temperature. For some assays, in vitro-stimulated cells were restimulated using PMA (50 ng/ml) and ionomycin (500 ng/ml) in the presence of brefeldin A (1 μg/ml) for 4 h prestaining for intracellular cytokines. Stained cells were analyzed using FACSCalibur (BD Biosciences) or LSR analyzer (BD Biosciences), and the data were analyzed using CellQuest (BD Biosciences), WinMDI (The Scripps Research Institute, La Jolla, CA), or Weasel (Walter and Eliza Hall Institute of Medical Research, Parkville, Victoria, Australia) applications. Specific regions were marked, and the gates and quadrants were set based on background staining using isotype control Ab for data analysis.

Cell-free supernatants, collected from 48-h T cell cultures, were tested for cytokines by ELISA or Luminex multiplex assays as per manufacturer’s directions (eBioscience, BD Pharmingen, R&D Systems, and/or BioSource International). The amount of cytokine was determined using an appropriate cytokine-specific standard curve. Background cytokine levels of effector cell cultures, in the absence of Ag, were subtracted from test values to calculate the actual cytokine response.

Total T cells and CD4⁺CD25⁺ cells were purified using magnetic sorting kits from Miltenyi Biotec, cultured in round-bottom 96-well plates for 48 h in the presence of anti-mouse CD3 Ab (2 μg/ml) and anti-mouse CD28 Ab (0.5 μg/ml), washed, and injected into 6-wk-old NOD-Scid mice individually and/or in combination. These T cell recipients along with uninjected control mice were tested for blood glucose levels every week. In a separate experiment, purified CD4⁺ T cells from 6-wk-old BDC2.5 TCR-Tg mice cultured in the presence of Ab-coated BDC peptide-pulsed DCs, as described above, with or without rIL-2 or TGF-β1 for 4 d. T cells isolated from these cultures (1 × 10⁶) were injected i.v. into 4-wk-old male NOD mice. Blood glucose levels in these mice were tested every other day. The animals with glucose levels >250 mg/dl for two consecutive bleeds were considered diabetic.

Pancreata were fixed in 10% formaldehyde and 5-μm paraffin sections were made and stained with H&E. Stained sections were analyzed in a blinded fashion using a grading system: 0, no evidence of infiltration; 1, peri-islet infiltration; 2, <25% infiltration; 3, 25–50% infiltration; and 4, >50% infiltration of each islet, as described in our earlier studies (19, 27, 28). Areas that appeared to have complete loss of islets were not included in this grading approach. About 100 islets were examined from each group.

Mean, SD, and statistical significance (p value) were calculated for this study. In most cases, values of test groups (Ag-pulsed anti–CTLA-4-Ab DCs) were compared with that of a control group (Ag-pulsed control-Ab DCs) unless specified otherwise. Because the same numbers of test and control values (data points) were compared, a paired two tailed t test was employed unless specified otherwise. Log-rank analysis was performed to compare diabetes-free mice in the test group with those in the control group. Fisher’s exact test was employed for comparing the total number of infiltrated islets in the test groups versus the control group. A p value ≤0.05 was considered significant.

Recently, we reported a method to enhance selective engagement of CTLA-4 on T cells from Ag-presenting DCs (16). In this DC-directed CTLA-4 engagement approach, we used BiAb (an agonistic anti–CTLA-4-Ab chemically cross-linked to a DC-specific anti-mouse CD11c-Ab)-coated DCs. Anti-CD11c Ab-linked anti–CTLA-4 Abs and control Abs were tested to confirm their ability to bind to BMDCs as shown in Fig. 1A. The ability of Ag-pulsed anti–CTLA-4-Ab DCs to activate T cells obtained from diabetic NOD mice in comparison with Ag-pulsed control-Ab DCs was tested in vitro. T cell proliferative response was significantly lower in cultures in which Ag-pulsed anti–CTLA-4-Ab DCs were used when compared with Ag-pulsed control-Ab DCs (Fig. 1B). As observed in Fig. 1C, Ag presentation by Ag-pulsed anti–CTLA-4-Ab DCs resulted in significantly lower IL-2 and IFN-γ responses by T cells from NOD mice. Suppressed proliferative, IL-2, and IFN-γ responses by T cells upon β cell Ag presentation by Ag-pulsed anti–CTLA-4-Ab DCs correlated with significantly enhanced IL-10 and TGF-β1 response by these T cells. These observations indicated enhanced CTLA-4–mediated negative signaling in T cells from NOD mice during pancreatic β cell Ag presentation by anti–CTLA-4-Ab DCs.

To examine whether the anti–CTLA-4 Ab DC-induced modulation of T cell response is the result of enhanced CTLA-4 engagement on T cells or due to some changes in the activation state of DCs upon Ab binding, Ab-coated as well as uncoated DCs were cultured for 36 h and examined for surface markers by FACS and secreted cytokines by ELISA. As observed in Fig. 1D and 1E, although both control-Ab and anti–CTLA-4-Ab DCs expressed relatively higher levels of CD40 and IL-6 compared with uncoated DCs, no difference was observed in the surface activation marker or secreted cytokine levels between control-Ab and anti–CTLA-4-Ab DCs. Of note, neither Ab-coated DCs nor uncoated DCs produced detectable amounts of IL-12, IL-10, TGF-β1, and IL-2 (not shown). These observations indicated that the ability of anti–CTLA-4-Ab DCs to modulate T cell response could be primarily due to enhanced CTLA-4 engagement and not a result of changes in the DC characteristics.

It is believed that pancreatic β cell Ag-specific T cells are primed primarily in the PnLNs, and reside in the pancreatic microenvironment (28–30). Therefore, we tested whether i.v.-transferred BMDCs can end up in the pancreatic microenvironment and secondary lymphoid tissues. Eight-week-old female NOD mice were injected with untreated DCs, Ag-pulsed anti–CTLA-4 DCs or Ag-pulsed control DCs. After 36 h, single-cell suspensions from the spleen, PnLNs, and pancreatic tissue of these mice were examined for CD11c⁺CFSE⁺ cells by FACS poststaining with PE-labeled anti-CD11c Ab. As shown in Fig. 2, a considerable number of CD11c⁺CFSE⁺ donor DCs were detected not only in spleen and PnLNs, but also in the pancreas of all three groups of DC-recipient mice. These results showed that the DC trafficking is not affected by Ab coating and suggested that anti–CTLA-4-Ab DCs might be able to present β cell Ag and engage CTLA-4 on T cells in secondary lymphoid organs as well as the target tissue.

To examine whether DC-directed CTLA-4 engagement approach that could suppress autoreactive T cells has an effect on T1D in NOD mice, prediabetic 8- or 12-wk-old female NOD mice were treated twice at 15 d apart with anti–CTLA-4-Ab or control-Ab DCs (5 × 10⁶ cells/mouse). Ag-pulsed anti–CTLA-4-Ab DCs produced a significant delay in hyperglycemia compared with control mice or Ag-pulsed control-Ab DC-recipient mice (Fig. 3A, 3B). Importantly, mice treated with Ag-pulsed control-Ab DCs developed hyperglycemia more rapidly compared with untreated control mice. Furthermore, anti–CTLA-4-Ab or control-Ab DCs that were not pulsed with Ag had no significant therapeutic effect even when the treatment was initiated at 8 wk of age. These results show that although Ag presentation by control mature DCs advances hyperglycemia presumably through induction and/or expansion of autoreactive T cells, anti–CTLA-4-Ab DCs delay hyperglycemia likely through suppression of these T cells. In addition, these results also indicated that the disease suppression observed in Ag-pulsed anti–CTLA-4-Ab DC recipients was most likely achieved through concurrent ligation of TCR and CTLA-4 on Ag-specific T cells. Of note, Ag-pulsed anti–CTLA-4-Ab DC treatment had no significant effect on hyperglycemia when the therapy was initiated at an early hyperglycemic or diabetic stage (not shown), perhaps due to irrevocable damage to islets at these later stages of the disease.

To examine whether treatment with Ag-pulsed anti–CTLA-4-Ab DCs has an effect on insulitis, pancreatic tissues from prediabetic NOD mice treated with Ag-pulsed anti–CTLA-4-Ab DCs and Ag-pulsed control-Ab DCs were examined for immune cell infiltration and islet damage. Sets of mice were euthanized 2 and 6 wk post second injection with DCs to obtain pancreatic tissues. As observed in Fig. 4, mice that received Ag-pulsed anti–CTLA-4-Ab DCs had significantly higher numbers of islets with less severe or no immune cell infiltration compared with control mice or Ag-pulsed control-Ab DC-recipient mice. Although >40% of islets in Ag-pulsed anti–CTLA-4-Ab DC recipients were insulitis free at 2 and 6 wk posttreatment, <20% of islets were immune cell infiltration free in Ag-pulsed control-Ab DC-recipient mice. On the other hand, whereas >55% of islets in Ag-pulsed control-Ab DC-recipient mice showed severe insulitis (grade ≥3), only <25% islets had severe immune cell infiltration in Ag-pulsed anti–CTLA-4-Ab DC recipients. In agreement with the disease data observed in Fig. 3, insulitis in Ag-pulsed control-Ab DC-recipient mice was relatively more severe than that noted in untreated control mice, indicating that treatment with peptide-pulsed mature DCs can aggravate the disease process. These results show that dominant selective coengagement of CTLA-4 along with β cell Ag presentation by DCs could lead to reduced infiltration and destruction of pancreatic islets by immune cells and delayed onset of hyperglycemia.

Next, we examined whether Ag-pulsed anti–CTLA-4-Ab DCs can modulate T cell response against β cell-specific Ag in vivo. Fifteen days post second injection with DCs, peptide-specific responses of T cells from DC recipients were compared. Ex vivo challenge of spleen and PnLN cells with β cell Ag peptides resulted in significantly higher TGF-β1 and IL-10 responses by T cells from Ag-pulsed anti–CTLA-4-Ab DC recipients compared with Ag-pulsed control-Ab DC recipients and nonrecipient control mice. In contrast, T cells from Ag-pulsed control DC recipients produced significantly higher IFN-γ response when compared with T cells from Ag-pulsed anti–CTLA-4-Ab DCs (Fig. 5A). In addition, PnLN T cells from Ag-pulsed anti–CTLA-4-Ab DC recipients produced relatively higher amounts of IL-4 compared with cells from Ag-pulsed control-Ab DC-recipient mice. However, IL-17 response was comparable in both control and anti–CTLA-4-Ab DC-recipient mice (not shown).

Examination of T cell proliferative response against β cell Ags by CFSE-dilution method revealed that a significant number of CD4⁺ T cells from Ag-pulsed control DC-recipient mice, but not T cells from Ag-pulsed anti–CTLA-4-Ab DC-recipient mice, proliferated in response to ex vivo peptide challenge (Fig. 5B). Importantly, a significantly higher number of T cells from Ag-pulsed control-Ab DC-recipient mice proliferated upon peptide stimulation compared with untreated mice. Similar T cell proliferative and cytokine responses were observed with cells obtained from control and test groups of mice 6 wk posttreatment (not shown). These observations suggested that although β cell Ag presentation by control-Ab DCs induced and/or expanded IFN-γ–producing autoreactive T cells, dominant engagement of CTLA-4 during Ag presentation by anti–CTLA-4-Ab DCs induced IL-10– and TGF-β1–producing hypoproliferative T cells.

Because T cells from Ag-pulsed anti–CTLA-4-Ab DC-treated mice were hypoproliferative when stimulated with β cell Ag peptides, we examined whether these T cells exhibit a Treg phenotype. Splenic and PnLN cells were harvested and tested for intracellular Foxp3 and IL-10 without additional peptide challenge ex vivo. As shown in Fig. 6A and 6B, both splenic and PnLN T cells from Ag-pulsed anti–CTLA-4-Ab DC-recipient mice showed significantly higher frequencies of Foxp3⁺ and IL-10⁺ T cell as compared with Ag-pulsed control-Ab DC recipients and nonrecipient control mice. Because CD4⁺ T cells from Ag-pulsed anti–CTLA-4-Ab DC recipients secreted significant amounts of TGF-β1 (Fig. 5A), cells from these mice were examined for surface-bound active and latent forms of TGF-β1. Unlike our recent study using Ag-immunized models (16), NOD mice did not show a significant number of T cells with surface-bound active or latent forms of TGF-β1 (not shown). These results indicate that suppression of insulitis and delayed onset of hyperglycemia observed in Ag-pulsed anti–CTLA-4-Ab DC-recipient NOD mice could be mediated by Foxp3-expressing and IL-10– and TGF-β1–secreting hypoproliferative T cells.

Because NOD mice that were treated with anti–CTLA-4-Ab DCs showed delayed hyperglycemia and increased frequencies of Foxp3⁺ and IL-10⁺ T cells compared with control-Ab DC-recipient mice, T cells from these mice were tested for their pathogenic nature, if any, and the ability to suppress diabetogenic cell (i.e., splenic T cells from hyperglycemic mouse) transfer-induced hyperglycemia in NOD-Scid mice. Fig. 7A shows that T cells from control-Ab DC recipients, but not anti–CTLA-4 Ab DC-recipient mice, induced hyperglycemia in NOD-Scid mice. More importantly, total T cells from anti–CTLA-4-Ab DC-recipient mice, but not control-Ab DC-recipient mice, induced significant protection against diabetogenic T cell-induced hyperglycemia. To examine the suppressor properties of Tregs from control-Ab and anti–CTLA-4-Ab DC-recipient mice, purified CD4⁺CD25⁺ T cells were adoptively transferred along with diabetogenic T cells into NOD-Scid mice. As observed in Fig. 7B, although CD4⁺CD25⁺ T cells from control-Ab DC-recipient mice delayed the onset of hyperglycemia, CD4⁺CD25⁺ T cells from anti–CTLA-4-Ab DC-recipient mice showed superior suppressor ability and prevented hyperglycemia in diabetogenic T cell-recipient mice. These observations showed that treatment of NOD mice with β cell Ag-pulsed anti–CTLA-4-Ab DCs results in the induction and/or expansion of T cells with regulatory properties.

As observed in Figs. 1 and 5, treatment of NOD mice with Ag-pulsed anti–CTLA-4-Ab DCs caused an increase in the frequencies of TGF-β1– and IL-10–producing T cells. We and others (12, 16, 31, 32) have shown that CTLA-4 engagement induces IL-10 and TGF-β1 production by T cells. A recent study (33) showed that CTLA-4–mediated suppression of T cell response results from downmodulation of CD28 expression on CD8⁺ T cells. Therefore, we examined how CD4⁺ T cell function is modulated using T cells from NOD-BDC2.5 TCR-Tg mice. As anticipated, presentation of BDC peptide by anti–CTLA-4-Ab DCs induced relatively lower proliferation of BDC2.5 T cells compared with control-Ab DCs (Fig. 8A). Further, consistent with the observation of Fig. 1, these T cells produced significantly lower IL-2 and IFN-γ but higher IL-10 and TGF-β1 compared with T cells that were cultured with Ag-pulsed control-Ab DCs (Fig. 8B). We examined the expression levels of costimulatory receptor CD28, activation marker CD69, and coinhibitor CTLA-4 on these T cells at different time points. As observed in Fig. 8C, no significant difference in the levels of CD28 expression was observed on T cells cultured with anti–CTLA-4-Ab or control-Ab DCs. Although levels of CTLA-4 and CD69 were comparable at early time points, expression levels of these markers were relatively lower on T cells cultured with Ag-pulsed anti–CTLA-4-Ab DCs at later time points, indicating that these cells are in a less activated state. This observation indicates that IL-10 and TGF-β1 produced upon enhanced CTLA-4 engagement may be responsible for the suppressed activation state and expression of relatively lower levels of CTLA-4 and CD69 on these T cells. These results also indicated that enhanced CTLA-4 engagement does not significantly affect CD28 expression on CD4⁺ T cells and suppression of T cell proliferation by anti–CTLA-4-Ab DCs may be cytokine dependent.

As observed in Fig. 8, T cells activated using anti–CTLA-4-Ab DCs showed significantly suppressed proliferative, IFN-γ, and IL-2 responses compared with those activated using control-Ab DCs. On the other hand, Ag presentation by anti–CTLA-4-Ab DCs resulted in IL-10 and TGF-β1 production by T cells. Therefore, we examined the role of IL-10 and TGF-β1 in enhanced CTLA-4 engagement-mediated modulation of T cell activation. Addition of IL-10– and TGF-β1–neutralizing Abs individually or in combination resulted in a significantly increased proliferative response by T cells in anti–CTLA-4-Ab DC-containing cultures (Fig. 9A). Moreover, IFN-γ and IL-2 levels were significantly increased in these cultures upon neutralization of IL-10 and/or TGF-β1 (Fig. 9B). These results suggest that IL-10 and TGF-β1 produced by T cells as a result of Ag presentation by anti–CTLA-4-Ab DCs are important in suppressed proliferation and IL-2 and IFN-γ production by T cells.

Anti–CTLA-4-Ab DCs could induce IL-10 and TGF-β1 production by T cells upon Ag presentation, and these regulatory cytokines are known to have the potential to promote adaptive Treg generation from effector T cells during activation. Therefore, Foxp3⁺ and IL-10⁺ T cell frequencies were examined in cultures in which anti–CTLA-4-Ab DCs were used for Ag presentation. Considerably higher numbers of T cells in anti–CTLA-4 Ab DC-containing cultures expressed Foxp3 and IL-10 compared with control-Ab DC-containing cultures (Fig. 9C). However, addition of TGF-β1–neutralizing Ab alone or along with anti–IL-10–neutralizing Ab negated this increase in Foxp3⁺ and IL-10⁺ T cell frequencies in anti–CTLA-4-Ab DC-containing cultures. These results indicated that TGF-β1 and IL-10 produced by T cells upon Ag presentation by anti–CTLA-4-Ab DCs can promote adaptive Treg induction and/or expansion.

The above-described observations and our previous report (16) showed that enhanced engagement of CTLA-4 upon presentation of Ag can result in the induction and/or expansion of T cells that express IL-10, TGF-β1, and Foxp3. Studies have shown that TGF-β1 is the critical cytokine known to promote adaptive Treg induction, and IL-2 supports this activity (34–37). In addition, CTLA-4 plays an important role in TGF-β1–mediated adaptive Treg induction (38). Although enhanced engagement of CTLA-4 on activated T cells induces TGF-β1 production, we and others (31, 39) have shown that CTLA-4 engagement can also suppress IL-2 production by these T cells. Therefore, whether TGF-β1 produced by T cells when stimulated with anti–CTLA-4-Ab DCs alone is sufficient to induce/expand T cells with regulatory properties was not known. In addition, what effect reduced levels of IL-2 may have on enhanced CTLA-4 engagement-associated expression of Foxp3, IL-10, and TGF-β1 by T cells was also not known. To examine these aspects, BDC2.5 TCR-Tg T cells were stimulated using BDC peptide-pulsed control-Ab or anti-CTLA-4-Ab DCs with exogenous TGF-β1 or IL-2 and examined for Foxp3 expression. As anticipated, addition of TGF-β1 resulted in an increase in the Foxp3⁺ T cell frequency in both Ag-pulsed control-Ab and anti–CTLA-4-Ab DC/T cell cultures compared with cultures that are not supplemented with exogenous cytokines (Fig. 10A). However, the increase in the Foxp3⁺ T cell frequency in the presence of exogenous TGF-β1 was profoundly higher in anti–CTLA-4-Ab DC/T cell cultures compared with control-Ab DC/T cell cultures. Of note, addition of rIL-10 did not result in an increase in Foxp3⁺ T cell frequency in these cultures (not shown). Interestingly, addition of rIL-2 also resulted in a significant increase in the frequency of Foxp3⁺ T cells in cultures in which Ag-pulsed anti–CTLA-4-Ab DCs, but not Ag-pulsed control-Ab DCs, were present. Similar to the observation shown in Fig. 9A, neutralization of TGF-β1 in rIL-2–supplemented cultures also reversed the Foxp3-inducing effect of Ag presentation by anti–CTLA-4-Ab DCs on T cells (Fig. 10A). These observations suggest that although TGF-β1 produced by T cells when the Ag is presented by anti–CTLA-4-Ab DCs can promote Foxp3⁺ induction, reduced IL-2 levels in the microenvironment can limit this property of TGF-β1. Importantly, a significantly higher number of T cells from both rIL-2– and rTGF-β1–supplemented Ag-pulsed anti–CTLA-4-Ab DC/T cell cultures produced IL-10 (Fig. 10B). As observed in Fig. 10C and 10D, addition of IL-2 increased CTLA-4 expression on T cells and elevated TGF-β1 production in the presence of Ag-pulsed anti–CTLA-4-Ab DCs, indicating a further enhancement of CTLA-4–mediated signaling in T cells of these cultures. These observations show that IL-2–induced upregulation of CTLA-4 on T cells likely facilitates selective enhanced ligation of this receptor by DC-bound anti–CTLA-4 Ab and triggers larger amounts of TGF-β1 secretion, leading to Foxp3 induction in an autocrine/paracrine manner.

Because a large number of BDC2.5 TCR-Tg T cells cultured with Ag-pulsed anti–CTLA-4-Ab DCs in the presence of rIL-2 or TGF-β1 expressed Foxp3 and IL-10, the diabetogenic property of these cells was examined in an adoptive transfer experiment. Many studies, including ours (27, 28), have used NOD-Scid mice in adoptive transfer experiments to examine the diabetogenic nature of T cells. However, considering the potential of lymphopenia-driven proliferation to affect the function of donor T cells in immune-deficient NOD-Scid mice, we used wild-type male NOD mice that are relatively less susceptible to spontaneous T1D compared with females for this study. Four-week-old male NOD mice were injected with BDC2.5 T cells that were cultured in the presence of Ag-pulsed control-Ab or anti–CTLA-4-Ab DCs with or without rIL-2 or rTGF-β1 and monitored for blood glucose levels. As observed in Fig. 11A, 100% of mice that received T cells stimulated with Ag-pulsed control-Ab DCs developed hyperglycemia within 6 d. In contrast, none of the mice that received T cells stimulated with Ag-pulsed anti–CTLA-4-Ab DCs developed hyperglycemia for at least 30 d. In addition, only the mice that received control-Ab DC-stimulated T cells, but not T cells from anti–CTLA-4-Ab DC cultures, showed a significant level of lymphocyte infiltration in the pancreatic islets (Fig. 11B). Similarly, mice that received T cells activated using control-Ab DCs, but not anti–CTLA-4-Ab DCs, in the presence of rIL-2 developed rapid hyperglycemia (Fig. 11C). However, mice that received T cells stimulated using Ag-pulsed control-Ab or anti-CTLA-4 Ab DCs in the presence of rTGF-β1 remained nondiabetic (Fig. 11D). These observations show that enhanced CTLA-4 engagement upon Ag presentation alters the diabetogenic property of activated BDC2.5 T cells and suggest that approaches to enhance CTLA-4 signaling could be employed for ex vivo modification of T cell function before their use for therapy.

In this study, we examined the potential of enhanced CTLA-4 ligation by β cell Ag-presenting DCs in modulating autoimmunity in a spontaneous model of T1D. We show that treatment using pancreatic β cell Ag-pulsed DCs coated with agonistic anti–CTLA-4-Ab can significantly delay T1D in NOD mice. This delay in hyperglycemia appears to be dependent on the suppressor function of Tregs induced and/or expanded upon enhanced engagement of CTLA-4 on effector T cells. Our observations demonstrate the potential of approaches to enhance the strength of CTLA-4–specific ligand on professional APCs to induce T cell tolerance and modulation of autoimmunity in T1D.

Susceptibility to T1D and other autoimmune diseases has been linked to CTLA-4 gene polymorphism (2, 4, 17). In addition, expression levels of soluble and ligand-independent forms of CTLA-4 have been implicated for the lack of peripheral tolerance to self-Ags (17). Previous studies have also suggested that T1D susceptibility is associated with an insufficient T cell regulation through CTLA-4 due to the expression of variants of CTLA-4, ligand strength on APCs, and defects in T cell activation (17, 18). In contrast, studies have shown that peripheral and ex vivo-generated DCs of patients with T1D and NOD mice are defective and, perhaps, deficient in their ability to provide effective help to natural Tregs (40–43). Our studies demonstrating that the natural Treg function can be enhanced in autoimmune models by modulating DC function supported this notion (43–46). Based on these observations, we hypothesized that enhancing the CTLA-4 engagement strength by increasing its selective ligand intensity on pancreatic β cell Ag-presenting APCs would have a modulatory effect on autoreactive T cells. Our recent study using immunization models showing robust adaptive Treg induction upon treatment with Ag-pulsed agonistic anti–CTLA-4-Ab DCs (16) further supported this notion. Therefore, to realize the clinical applicability of this approach, we examined whether enhancing the CTLA-4–specific ligand strength on pancreatic β cell Ag-presenting DCs using an agonistic Ab can modulate spontaneously occurring autoimmunity and T1D in NOD mice.

Despite the widely reported abnormalities in DC function and CTLA-4 signaling in NOD mice, our observations show that Ag presentation by anti–CTLA-4 Ab-coated NOD mouse DCs can suppress the proliferation of β cell Ag-specific T cells from the spleen cells of diabetic mice. Enhanced CTLA-4 engagement during β cell Ag presentation not only resulted in the production of significant amounts of suppressor cytokines, such as of IL-10 and TGF-β1, but also suppressed IL-2 and IFN-γ responses. These findings indicated that the T cells are activated upon Ag recognition, but the cytokine profile is skewed as a result of enhanced CTLA-4 engagement. Importantly, activation of T cells by natural ligands expressed on Ag-presenting DCs is critical for providing a β cell Ag-specific signal to achieve the primary goal of selectively modulating autoreactive T cell function.

Although several immunodominant self-antigenic peptides are identified in T1D, it is widely believed that autoimmune response in T1D might be directed against a wide array of β cell-associated proteins. Therefore, use of one or a few known antigenic peptides for controlling T1D may have a limited therapeutic value. Nevertheless, the ability of Ag-pulsed anti–CTLA-4 DCs to induce significant amounts of suppressor cytokines and concomitantly suppress effector cytokine production in T cells from diabetic mice in vitro prompted us to examine whether these DCs can delay the onset and/or treat hyperglycemia in prediabetic and diabetic NOD mice. A significant delay in the onset of hyperglycemia observed in prediabetic mice that received Ag-pulsed anti–CTLA-4-Ab DCs suggests that the autoimmune response is suppressed in these mice, which prolongs the normal functioning of remaining β cells. However, suppression of autoimmunity using Ag-pulsed anti–CTLA-4-Ab DCs alone does not appear to be sufficient to reverse established hyperglycemia, perhaps due to the loss of a majority of the functional islets prior to the initiation of the treatment. Combinational approaches, such as therapy to enhance β cell mass and function along with DC-directed CTLA-4 engagement, might hold the potential for reversing already established hyperglycemia.

Relentless destruction of β cells through apoptosis and sustained endogenous Ag presentation in the pancreatic microenvironment leads to continued β cell destruction under inflammatory conditions and contributes to the progression of T1D (47, 48). Therefore, it is possible that β cell Ags are continuously released from dying or dead cells. If sufficient amounts of β cell Ags are released, then treatment with DCs that are not loaded with antigenic peptide might also be able to produce a similar protective effect by capturing, processing, and presenting endogenous β cell Ags to T cells. However, our observations show that β cell Ag-pulsed DCs, but not DCs that were not pulsed with Ag, modulated the disease outcome (either aggravation or suppression), indicating that mature DCs do not capture sufficient amounts of β cell Ags in vivo and present to pathogenic T cells. Therefore, loading the DCs with antigenic peptide prior to anti–CTLA-4 Ab coating and injection is critical for achieving a significant therapeutic effect.

T cells from NOD mice treated with Ag-pulsed anti–CTLA-4-Ab DCs produced significant amounts of IL-10 and TGF-β1, suggesting that these cytokines and T cells may be responsible for protecting remaining functional β cells in prediabetic mice. These T cells appeared to be hypoproliferative in nature, as indicated by their inability to proliferate significantly upon challenge with β cell antigenic peptides. Although the suppressor cytokine response by T cells exposed to Ag-pulsed anti–CTLA-4-Ab DCs was significantly higher, the effector cytokine (IFN-γ) response was lower when compared with T cells from control-Ab DC recipients. This suggests that enhanced CTLA-4 engagement upon Ag presentation can skew the T cell response from pathogenic toward suppressor type.

The DCs are recognized as the only type of APCs that are capable of activating naive T cells against a particular Ag. Although upon inoculation Ag-pulsed anti–CTLA-4-Ab DCs may primarily target existing memory T cells, it is possible that they will also activate naive T cells through β cell Ag presentation. Therefore, we assume that although Ag-pulsed control-Ab DCs can induce new pathogenic T cells from naive cells along with expansion of existing memory T cells, Ag-pulsed anti–CTLA-4-Ab DCs may induce and/or expand Ag-specific T cells with suppressor phenotype. This may partially explain why a significantly higher number of T cells from control DC-recipient mice compared with untreated control mice proliferated upon challenge with β cell Ag. More severe insulitis in Ag-pulsed control-Ab DC-recipient mice compared with untreated mice is also indicative of this effect. This is in contrast to the induction of hypoproliferative T cells with suppressor phenotype that could delay the progression of insulitis upon treatment with Ag-pulsed anti–CTLA-4-Ab DCs.

Our earlier study using animals immunized with foreign or self-Ag has shown that profound numbers of adaptive Tregs with Foxp3 expression and surface-bound TGF-β1 are induced upon treatment using cognate Ag-pulsed anti–CTLA-4-Ab DCs (16). In contrast, the current study using a spontaneous autoimmune diabetic model showed no significant difference in the frequency of T cells with surface-bound TGF-β1 in test and control groups of mice. Further, Tregs induced and/or expanded in the spontaneous NOD model does not appear to be as robust as those we have found in the immunization-induced disease model. This difference might be attributable to relatively fewer Ag-specific T cells in the former model. Further, in the current study, we used only a limited number of self-antigenic peptides, and, therefore, T cells with specificities toward these peptides, which may represent only a small proportion of the pathogenic T cell repertoire present in this spontaneous disease model, were affected. Yet another explanation for the difference in the frequencies of Tregs in an immunization model versus a spontaneous model upon treatment using Ag-pulsed anti–CTLA-4-Ab DCs could be the difference in the amount of IL-2 and/or TGF-β1 produced in the lymphoid microenvironment during enhanced CTLA-4 engagement. As suggested by our in vitro experiments, both IL-2 and TGF-β1 can promote enhanced CTLA-4 engagement-mediated induction of T cells with regulatory properties. Because the overall levels of IL-2 and TGF-β1 in an immunization model are expected to be higher due to the presence of large numbers of Ag-specific T cells compared with a spontaneous model, we believe that these cytokines may be contributing to a robust Treg response in the former model.

Irrespective of the above-described difference in the Treg frequencies, the current study shows relatively higher frequencies of Foxp3⁺ and IL-10⁺ T cells in freshly isolated spleen and PnLNs of Ag-pulsed anti–CTLA-4-Ab DC-recipient NOD mice compared with Ag-pulsed control-Ab DC recipients. In addition, as observed in our earlier study, T cells from Ag-pulsed anti–CTLA-4-Ab DC-recipient mice secreted higher amounts of IL-10 and TGF-β1 upon challenge with Ag. This suggested that enhancing CTLA-4–specific ligand strength on DCs could be an effective way of promoting self-Ag–specific tolerance.

Although self-Ag–specific T cell tolerance is the key feature of our DC-directed CTLA-4 engagement approach, we could not achieve a lasting protection from diabetes in NOD mice using a short-term treatment that targeted a limited repertoire of T cells with known Ag specificity. Failure to achieve long-term benefit may be due to the inherent defects in the NOD immune system, repertoire spreading to include other peptide specificities, and regeneration and/or reactivation of autoreactive T cells. Targeting a significant portion of autoreactive T cells by pulsing the DCs with a mixture of full-length pancreatic β cell Ags instead of selected peptides and/or intermittent injections with these therapeutic DCs may prove effective in achieving a lasting protection from T1D. Importantly, a robust increase in functional Treg frequency that may produce long-lasting protection from the disease in anti–CTLA-4-Ab DC-treated mice is hindered by enhanced CTLA-4 signaling-mediated suppression of IL-2 secretion by T cells. Although enhanced CTLA-4 engagement upon Ag presentation can result in the production of significant amounts of IL-10 and TGF-β1 by T cells, this cytokine milieu alone does not appear to be sufficient to promote a robust Treg response, even with a suppressed level of inflammatory cytokines, such as IFN-γ. Therefore, based on our observations showing the ability of exogenous IL-2 to promote a profound increase in Foxp3⁺ and IL-10⁺ T cell frequencies in anti–CTLA-4-Ab DC-containing cultures, we believe that therapeutic efficacy of DC-directed CTLA-4 engagement approach could be profoundly enhanced by coadministration of rIL-2. Future studies are important and required to address this notion.

Approaches to modulate costimulatory and coinhibitory functions are considered very effective in inducing T cell tolerance for treating autoimmunity and transplant rejection. Although blockade of these pathways has shown therapeutic potential in various conditions, methods to enhance signaling through the dominant T cell repressor-receptor concurrent with TCR engagement has excellent therapeutic potential due to the active nature of T cell downregulation. In fact, enhancing CTLA-4–specific ligand strength on APCs and target cells aimed at downregulating T cell response has been tested in both allotransplant and autoimmune models (11, 16). We have reported that thyroid-targeted delivery of anti–CTLA-4 Ab leads to suppression of the antithyroglobulin immune response and suppression of experimental autoimmune thyroiditis (11). We and others have also shown that coating allogeneic cells with anti–CTLA-4 Ab or transfecting these cells to express single-chain anti–CTLA-4 Ab on the surface could substantially reduce the immune response against that alloantigen in the recipient mice and induce adaptive Tregs with the ability to produce IL-10 and/or TGF-β1 (12–14, 16). In addition, Tg expression of anti-CTLA-4 agonistic Ab on B cells results in delayed hyperglycemia in NOD mice (15). Similarly, coadministration of vector constructs encoding a CTLA-4–specific ligand, B7.1wa, and an islet-specific protein in a DNA vaccination study has demonstrated the ability to delay hyperglycemia in NOD mice (49). Our recent study showing that a robust adaptive Treg response can be induced using Ag-pulsed anti–CTLA-4-Ab DCs in Ag-primed mice (16) indicated that enhancing CTLA-4–specific ligand strength on APCs, DCs in particular, is an effective strategy for treating autoimmunity. Importantly, defective DC function, CTLA-4 signaling, and the spontaneous onset of the disease, as in human patients with T1D (17, 18, 41–43), make the NOD mouse model of T1D a unique system to demonstrate the therapeutic efficacy of this DC-directed CTLA-4 engagement approach. This study also provides additional insights on the mechanism of enhanced CTLA-4 engagement-mediated adaptive Treg induction and/or expansion and suggests methods to enhance the therapeutic efficacy of this approach.

Disclosures The authors have no financial conflicts of interest.