Enhanced Engagement of CTLA-4 Induces Antigen-Specific CD4⁺CD25⁺Foxp3⁺ and CD4⁺CD25⁻ TGF-β1⁺ Adaptive Regulatory T Cells¹

Positive and negative coreceptors of the CD28 family have powerful modulatory effects on the activation, proliferation, and cytokine production of T cells. T cell activation requires both Ag-specific triggering of the TCR complex and costimulation mediated through the positive coreceptor CD28 (1, 2). Ligation of CD28 on T cells by its ligands CD80 (B7.1) and CD86 (B7.2), which are primarily found on APCs, leads to IL-2 production and T cell proliferation. The negative coreceptor, CTLA-4, 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 (3).

CTLA-4 can mediate negative regulation of T cell responses through several mechanisms. Although its competition with CD28 for CD80/86 binding results in lowered activation, signaling through the cytoplasmic tail of CTLA-4 and modulation of TCR signaling by activating phosphatases such as SHP-2 leads to cell cycle disruption and suppression of IL-2 production. Recently, it has been shown that the binding of CTLA-4 to CD80 and CD86 can lead to modulation of dendritic cell (DC)³ function by inducing the release of IDO (4, 5). CTLA-4 mediated interference of lipid raft formation on the T cell surface and disruption of CD28 localization at the immunological synapse have also been implicated in the suppression of T cell response (6, 7). Recently, it was shown that CTLA-4 can reduce the contact period between T cells and APCs and lead to a decrease in proinflammatory cytokine production and proliferation (8). These properties indicate a pivotal role for CTLA-4 in regulating T cell function. Therefore, manipulating CTLA-4 signaling could be an effective strategy to modulate the immune response for treating various immune-mediated clinical conditions (9, 10, 11, 12, 13).

While the blockade of CTLA-4 signaling using an anti-CTLA-4 Ab resulted in protection against tumors and viral and bacterial infections, the blockade of CD28 signaling using CTLA-4-Ig has shown promise in treating autoimmunity and transplant rejection (9, 10, 12, 13). Although a role for CTLA-4 in the negative regulation of T cells is well recognized, its contribution to Treg function remains controversial (14, 15, 16, 17). More importantly, although the requirement of CTLA-4 in TGF-β1-mediated adaptive Treg generation has been reported (18), a direct role for CTLA-4 signaling in Treg induction is not known. Earlier, we showed that coating an autoimmune or alloimmune target with anti-CTLA-4 agonistic Ab can result in an increase in the memory regulatory T cell (Treg) numbers in vivo (19, 20). However, the direct effect of enhanced CTLA-4 signaling in the context of Ag-specific TCR-engagement on T cell differentiation into adaptive Tregs has not been understood.

The ability of B7.1 and B7.2 to bind to CD28 has precluded examination of the outcome of selective engagement of CTLA-4 in vivo. In this study, we examined the effect of dominant CTLA-4 engagement by Ag-presenting DCs coated with an agonistic Ab on T cell response and differentiation in vivo and found that it can result in the induction of Ag-specific adaptive Tregs. Our results also demonstrate that CTLA-4 engagement by DCs coated with an agonistic anti-CTLA-4 Ab can produce long-lasting protection from autoimmunity through the suppression of thyroglobulin-specific T cell and Ab responses.

Six- to 8-wk-old female wild-type BALB/c, MHC class II OVA_(323–339) epitope-specific TCR-transgenic (D011.10 in BALB/c background), and wild-type CBA/J mice were purchased from The Jackson Laboratory. All animal studies were approved by the animal care and use committee of the University of Illinois, Chicago IL.

Hamster anti-mouse CTLA-4 (clone UCI0-4-F-I0-11) and CD11c (clone N418) hybridomas were purchased from American Type Culture Collection and grown in serum-free/protein-free medium (BD Biosciences) and the Abs were purified using protein L or A (Sigma-Aldrich) columns. Anti-CTLA-4-anti-CD11c bispecific Ab (BiAb) was prepared and purified as described earlier (19, 20, 21). Purified hamster IgG (Fitzgerald International) linked to the anti-CD11c Ab served as a control BiAb. Ag binding efficiencies of BiAbs were tested by FACS using bone marrow (BM)-derived DCs (BMDCs) and ELISA using recombinant CTLA-4-Ig (R&D Systems) as described earlier (21).

Type VI OVA and hen egg lysozyme (HEL), LPS from Salmonella enterica, and CFA were purchased from Sigma-Aldrich. OVA_(323–339) peptide was synthesized in the Research Resources Center at the University of Illinois (Chicago, IL). Mouse thyroglobulin (mTg) was prepared as described earlier (19, 22). Purified anti-mouse-TGF-β1 (clone A75-2; nonneutralizing) and anti-CD16/CD32 (Fc block) Abs were purchased from Caltag Laboratories. FITC-, PE-, PE-Cy5- and PE-Texas Red-conjugated Abs were from BD Pharmingen, eBioscience, R&D Systems, and Biolegend Laboratories. Paired Abs and the required cytokine standards for detecting mouse IL-2, IL-4, IFN-γ, and IL-10 (eBioscience) and TGF-β1 (R&D Systems or BD Pharmingen) were used in ELISA. Neutralizing Ab to mouse IL-10 (clone JES5-2A5) was purchased from eBioscience. Recombinant mouse IL-2, neutralizing Ab to mouse TGF-β1 (clone 1D11), and normal rat IgG1 Ab were purchased from R&D Systems. Magnetic bead-based cell isolation kits were purchased from Miltenyi Biotec. Multiplex cytokine assay reagents were purchased from BioSource International.

Wild-type BALB/c mice were i.v. injected with 50 μg of OVA or HEL and 5 μg LPS at least 10 days before being used in experiments. In some experiments, OVA was emulsified in CFA and s.c. injected. Cells from draining lymph nodes (LNs) were obtained from these mice 15 days after priming.

DCs were either isolated from the spleens of wild-type BALB/c mice or generated in vitro from BM cells. DCs from spleens were isolated using magnetic bead labeled anti-CD11c Ab (Miltenyi Biotec) according to the manufacturer’s directions. The percentage of CD11c⁺ cells in the enriched population was generally >90%. For generating DCs in vitro, BM cells were cultured in complete RPMI 1640 medium containing 10% heat-inactivated FBS in the presence of 20 ng/ml GM-CSF at 37°C in 5% CO₂ for 2 days and then for a further 4 days in fresh complete RPMI 1640 medium containing 20 ng/ml GM-CSF and 5 ng/ml IL-4. The nonadherent cells from 6-day cultures were used.

Before being used in the experiments, DCs (1 × 10⁶/ml) were incubated for 48 h at 37°C in the presence of OVA or mTg (20 μg/ml), or OVA peptide (2 μg/ml) with or without LPS (5 μg/ml), 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 Ab-coated DCs. Bound Ab levels on DCs were tested by FACS analysis before every experiment after staining with FITC-labeled anti-hamster IgG Abs. Maturation status of LPS-treated DCs was also tested in comparison with untreated DCs by using Abs against the activation markers CD80, CD86, CD40, and MHC II by FACS before every experiment.

Control and test BiAbs were tested for their ability to bind to respective Ags and stay on the DCs upon coating. ELISA was conducted to test the binding efficiency of the CTLA-4 portion of the BiAb to recombinant CTLA-4-Ig as described earlier (19, 20). Binding efficacy of anti-CD11c portion of the BiAb to CD11c expressed on DCs was tested by FACS. The persistence of BiAb on coated DCs was tested by analyzing aliquots of cells obtained at different time points in a FACSCalibur flow cytometer (BD Biosciences). The persistence of Abs on a DC surface was also tested using a Zeiss LSM 510 confocal microscope.

Ag-pulsed, Ab-coated DCs (1 × 10⁵ cells/well) were plated in 96-well flat-bottom tissue culture plates in triplicate along with purified T cells (5 × 10⁵ cells/well) from OVA-primed mice in RPMI 1640 medium containing 2% mouse serum. After 48 h, cells were pulsed with l μCi/well [³H]thymidine for 18 h. Thymidine incorporation was measured as described previously (19, 20).

OVA-primed mice were injected i.v. with 5 × 10⁶ OVA or OVA-peptide pulsed DCs and control or anti-CTLA-4 Ab-coated mature DCs once or twice at a 10-day interval and sacrificed on day 15 postinjection to test for T cell response to ex vivo challenge with the Ag. To test the Ag specificity of T cell suppression, HEL-primed mice were adoptively transferred (i.v.) with CD4⁺ T cells from DO11.10 TCR-transgenic mice, followed by OVA_323–339 peptide-pulsed and Ab-coated DCs (5 × 10⁶ cells/mouse) after 24 h. These mice were sacrificed on day 15 posttreatment to test for Ag-specific T cell responses.

T cell proliferation against ex vivo antigenic challenge (10 μg/ml OVA, 10 μg/ml HEL, 1 μg/ml OVA peptide, and 20 μg/ml mTg) was tested either by [³H]thymidine incorporation as described above or a CFSE dilution assay as described earlier (20). CFSE dilution was measured by FACS analysis after 5 days of culture. In some assays, saturating concentrations of neutralizing anti-mouse IL-10 (1 μg/ml), anti-mouse TGF-β1 (1 μg/ml), and/or isotype-matched control Abs were added. In some assays, enriched T cell subpopulations were cultured in the presence of OVA (50 μg/ml) or anti-CD3 and CD28 Abs (5 μg/ml each) in the presence or absence of rIL-2 (50 U/ml).

Cells were washed with PBS supplemented with 2% FBS (pH 7.4) and blocked with anti-CD16/CD32 Fc block Ab on ice for 15 min. For surface staining, cells were incubated with FITC-, PE-, and PE-Cy5- or PE-Texas Red-labeled appropriate Abs in different combinations on ice for 30 min. For intracellular staining, surface-stained cells were fixed, permeabilized using fixation/permeabilization kits (Ebioscience), and incubated with fluorochrome-labeled Abs. Stained cells were analyzed using a FACSCalibur, LSR, or CyAn analyzer, and the data were analyzed using the CellQuest, WinMDI, Weasel, or Summit applications. Specific regions were marked, and the gates and quadrants were set while analyzing the data based on isotype control Ab background staining.

Cell-free supernatants collected after 48 h were tested for cytokine levels by ELISA or Luminex multiplex assays as per the manufacturer’s directions (BD Pharmingen and 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.

CD4⁺ and CD4⁺CD25⁺ T cell subpopulations were isolated using the appropriate kits and magnetic separation columns (Miltenyi Biotec). T cell subpopulations were also purified by high speed FACS. Cells were stained with PE-Texas Red-anti-CD4, PE-Cy5-anti-CD62L, PE-anti-CD25 (clone 74D; noncytotoxic IgM Ab), and anti-TGF-β1-biotin/streptavidin-FITC (clone A75-3; nonneutralizing), and the CD4⁺CD62L^highCD25⁺, CD4⁺CD62L^lowCD25⁺, CD4⁺CD62L^highTGF-β1⁺, CD4⁺CD62L^lowTGF-β1⁺, and CD4⁺CD25⁻TGF-β1⁻ subpopulations were sorted using the MoFlo high speed sorter (DakoCytomation).

Effector T cells (CFSE labeled) from OVA-primed mice were mixed with enriched subpopulations of T cells from tolerant mice at various ratios. These mixtures or individual cell populations were used in T cell proliferation assays (total of 0.6 × 10⁵ cells/well) either in the presence or absence of OVA. For Transwell assays (BD Biosciences), Treg subpopulations from tolerant mice and APCs in the well insert (upper compartment) and CFSE-labeled T cells from OVA-primed mice and APCs (2.5 × 10⁶) in the well (lower compartment) in a 24-well plate were cultured in the presence or absence of OVA. CFSE dilution was measured by FACS analysis after 5 days of incubation.

For inducing experimental autoimmune thyroiditis, 8-wk-old CBA/J mice were i.v. injected with 100 μg of mTg along with 25 μg of bacterial LPS on day 0 and 100 μg of mTg and 5 μg of LPS on day 10. Mice were treated with mTg-pulsed Ab-coated DCs on days 15 and 25 and sacrificed on day 40 for disease evaluation. In some experiments, mice were rechallenged with 100 μg of mTg and 5 μg LPS on day 85 and sacrificed on day 100. Thyroids were fixed and 5-μm paraffin sections were made and stained with H&E to examine lymphocyte infiltration. Cellular infiltration and follicular damage were scored as grades 0–4 as described earlier (19, 22). mTg-specific T cell proliferative and cytokine responses were tested using spleen and LN cells, and serum samples were tested for mTg-specific Ab response by ELISA as described earlier (19, 22, 23).

Mean, SD, and statistical significance (p value) were calculated using a Microsoft Excel or SSPS statistical application. In most cases, values of the test group (mice that received anti-CTLA-4 Ab-coated DCs) were compared with that of control group (mice that received isotype control Ab-coated DCs) unless specified. P ≤ 0.05 was considered significant.

To generate an effective method for enhanced CTLA-4 engagement upon Ag presentation, we adopted a BiAb approach that we have thoroughly tested in our earlier studies (19, 20, 21). In this study, agonistic anti-CTLA-4 Ab (as test) or purified hamster IgG (as control) was chemically linked to DC-specific anti-mouse CD11c Ab. Because both anti-CD11c and anti-CTLA-4 Abs were of hamster origin and there is no reliable isotype-specific anti-hamster Ab available, we adopted the following approaches to test the cross-linked Ab. We confirmed the efficiency of the BiAb to bind to CTLA-4 and CD11c by using CTLA-4 Ig as the Ag in an ELISA and BMDCs as CD11c-expressing cells in a FACS assay (not shown). As expected, confocal microscopy and FACS analyses showed that mature DCs are better able to retain Abs on the surface compared with immature DCs (Fig. 1). Further characterization indicated that presence of optimum amounts of Abs (1 × 10⁷ DCs coated with ≤10 μg of BiAb in 1 ml) on the DC surface has no significant effect on their Ag-presenting function (not shown). Therefore, 10 μg BiAb was used for coating every 1 × 10⁷ DCs in a 1-ml volume throughout this study.

The effect of Ag-pulsed, anti-CTLA-4 Ab-coated DCs on T cell activation was tested in vitro using naive and Ag-primed CD4⁺ T cells. Anti-CTLA-4 Ab-coated DCs suppressed Ag-exposed but not naive T cell responses (Fig. 2,A). This suggested either a lack of or slower recruitment of CTLA-4 on these T cells compared with Ag-primed T cells. To examine whether anti-CTLA-4 Ab-coated untreated and LPS-treated DCs can suppress T cell response, OVA-pulsed Ab-coated DCs were incubated with CD4⁺ T cells from OVA-primed mice. As anticipated, LPS-treated DCs induced significantly higher activation of T cells compared with untreated DCs (Fig. 2,B). Although the early activation of T cell response was not significantly different, anti-CTLA-4 Ab-coated, LPS-treated DCs induced a significant suppression of T cell proliferation and IL-2 production and higher amounts of IL-10 and TGF-β1 production compared with control Ab-coated DCs, indicating a dominant CTLA-4-mediated signaling (Fig. 2 C). This suggests that the initial activation, an essential step required for the up-regulation of CTLA-4 on T cells, is unaffected by the presence of agonistic anti-CTLA-4 Ab on the DC surface but that the subsequent enhanced engagement of CTLA-4 efficiently reduces the proliferation of these T cells. In addition, the suppression of T cell proliferative response was induced only by anti-CTLA-4 Ab-coated and LPS-treated, and not by untreated, DCs. This could be due to the superior ability of LPS-treated DCs to induce T cell activation and up-regulation of CTLA-4 and to maintain sufficient amounts of Ab on the surface for a longer period for an effective enhanced engagement of CTLA-4. Therefore, LPS-treated DCs were used for rest of the study.

To examine the effect of treatment with Ag-pulsed, anti-CTLA-4 Ab-coated DCs on T cell activation in vivo, OVA-primed mice were treated i.v. with OVA-pulsed DCs that were coated with either hamster IgG (control group) or anti-CTLA-4 Ab (test group). CD4⁺ T cells from these mice were tested for surface expression of the Treg marker CD25 and the memory-related marker CD62L and their response to ex vivo challenge with OVA. As seen in Fig. 3,A, although the test group of mice demonstrated a significant increase in CD4⁺CD25⁺ T cell numbers compared with control mice (p = 0.032), they appeared to have relatively lower, although not statistically significant, CD62L^low memory T cell numbers. To examine the effect of challenge exposure to OVA, these T cells were maintained in the presence of OVA ex vivo and tested for the early activation marker CD69 after 16 h, for cytokines after 48 h, and for proliferation after 5 days. The numbers of cells expressing the early activation marker CD69 were not significantly different in the test and control groups. Importantly, CD4⁺ T cells from the test group of mice showed a significantly lower number of T cells proliferating compared with control mice (7.19 vs 20.14%) (Fig. 3,B). In addition, as indicated by the CFSE dilution, T cells from the test group of mice showed profoundly slower proliferation and/or less number of divisions compared with the control T cells. Consistent with the proliferation response, T cells from the test group of mice produced significantly less IL-2 and IFN-γ compared with cells from control mice (p < 0.013) (Fig. 3,C, left panels). However, cells from these mice produced considerably higher IL-10 and TGF-β1 than the controls (p < 0.0094) (Fig. 3 C, right panels). Collectively, these results suggest that DC-directed, enhanced CTLA-4 engagement during Ag presentation resulted in the induction of Ag-specific hyporesponsive T cells. As evident from CD69, IL-10, and TGF-β1 expressions, these T cells are, in fact, responsive to cognate Ag but hyporesponsive in terms of their proliferative and proinflammatory cytokine responses. This observation and the presence of a higher number of CD25⁺ T cells further suggest the presence of Ag-specific regulatory/suppressor T cells in the test group of mice and that these cells can be functionally activated upon challenge with the Ag.

Because the test group of mice showed significantly higher numbers of CD4⁺CD25⁺ T cells (Fig. 3,A) and the frequency was boosted by an additional dose of DCs (not shown), we speculated that these CD4⁺CD25⁺ T cells may have a role in the Ag-specific hypoproliferative response of effector T cells in the test mice. To examine this possibility, total CD4⁺ and CD4⁺CD25⁻ T cells from both test and control mice were isolated (Fig. 4,A) and tested for their proliferative responses against OVA. T cells from both test and control mice showed a small increase in the proliferative response to OVA challenge when CD4⁺CD25⁺ T cells were depleted (Fig. 4,B, upper two panels). Although CD4⁺ T cells from the test group of mice showed a significant reduction in TGF-β1 and IL-10 responses when CD4⁺CD25⁺ cells were depleted (p < 0.0021), a noticeable change in the level of IL-10, but not TGF-β1, was observed with control T cells (Fig. 4,C, right panel). In a recall response assay, neutralization of TGF-β1 alone or together with IL-10 caused a profound increase in T cell proliferative (23.8 and 26.1%, bottom panel, compared with 11.9%, middle panel of Fig. 4,B) and inflammatory cytokine (i.e., IFN-γ and IL-2; left panels of Fig. 4 C) responses in cells from the test group of mice.

To further examine the role of CD4⁺CD25⁺ and CD4+CD25⁻ T cells from the Ab-coated DC-treated mice, they were cocultured with CFSE-labeled, purified effector T cells from OVA-primed mice along with Ag-pulsed APCs. Ag-specific T cell proliferative response was significantly suppressed by both CD4⁺CD25⁺ and CD4⁺CD25⁻ subpopulations of T cells, with a more potent inhibition of proliferation by the enriched CD4⁺CD25⁺ population from the test group (12.4 vs 22.2% of cells with CFSE dilution) at a 2:1 effector:suppressor ratio. (Fig. 4,D, upper panel). Further, the effector T cell suppression induced by these subpopulations was also higher in terms of their rate of proliferation (mean fluorescence intensity (MFI) of 482 and 592 vs 325 in the absence of suppressor cells). Importantly, only the CD4⁺CD25⁺ population from the control group of mice demonstrated a considerable ability to suppress the effector T cell response, albeit significantly lower compared with that of CD4⁺CD25⁺ cells from the test group. When the assay was performed in a Transwell system (Fig. 4 D, lower panel), CD4⁺CD25⁻ T cells from the test group of mice showed no significant suppression of effector T cell proliferation. However, CD4⁺CD25⁺ cells from these mice retained some ability to suppress Ag-specific T cell proliferation (CFSE dilution in 22.8% of cells compared with 31.9% in the culture with no suppressor cells), albeit lower compared with when the cells were in the same chamber. This suggested that the effector T cell suppression by CD4⁺CD25⁻ cells is contact dependent, whereas suppression by CD4⁺CD25⁺ cells is most likely through secreted cytokines and the effect is probably diluted when the suppressors are not in close proximity to the effectors.

The above described results demonstrated three important aspects of T cell hyporesponsiveness in anti-CTLA-4 Ab-coated DC treated mice. These are: 1) an incomplete restoration of Ag-specific T cell response upon depletion of CD4⁺CD25⁺ T cells; 2) the ability of CD4⁺CD25⁻ T cells to suppress effector T cells in a contact-dependent manner; and 3) the reversal of the hypoproliferative response of T cells from the test group of mice by neutralizing anti-TGF-β1 Ab. These observations prompted us to test CD4⁺ T cell populations from treated mice for surface-bound TGF-β1. T cells from the test mice showed a significantly higher number of TGF-β1⁺ T cells compared with control and untreated naive mice. Importantly, depletion of CD4⁺CD25⁺ T cells (middle panel of Fig. 5,A) did not significantly affect the overall percentage of CD4⁺TGF-β1⁺ T cells in the test group (lower panel of Fig. 5 A and also B). This suggested that the regulatory function of CD4⁺CD25⁻ T cells from the test group is mediated by cells with surface-bound TGF-β1⁺ (CD4⁺CD25⁻TGF-β1⁺) T cells.

Because the CD62L^low memory T cell levels were relatively lower in the test mice (Fig. 3,A), we examined them for memory and naive subpopulations of Tregs. Although the total number of CD4⁺ T cells with a memory phenotype (CD62L^low) in the test group was lower compared with that of the control group (47.19 vs 54.14%) even after a second dose of DC treatment (Fig. 6,A), ∼150% more memory CD4⁺CD25⁺ (CD4⁺CD25⁺CD62L^low) cells were detected in test mice (12.57 vs 26.02%). In contrast, the percentage of naive CD4⁺CD25⁺ T cells (CD4⁺CD25⁺CD62L^high) was similar in both groups (10.36% and 11.89%) (Figs. 6,B and 4,C). Intracellular staining revealed that a majority of the CD4⁺CD25⁺CD62L^low cells from the test group were foxhead box p3 transcription factor (Foxp3) positive (Fig. 6,D). However, part of the CD4⁺CD25⁺CD62L^low T cells from the control group were Foxp3⁻, indicating that a significant number of these T cells are perhaps in a state of activation. Further, CD4⁺CD25⁺CD62L^low cells from the test group showed relatively higher CTLA-4, TGF-β1, IL-10, and glucocorticoid-induced tumor necrosis factor receptor (GITR) expression compared with CD4⁺CD25⁺CD62L^high cells (Fig. 6 E). Whereas CD4⁺CD25⁺CD62L^high cells from control mice showed expression of these markers at levels comparable to that of the test group, control CD4⁺CD25⁺CD62L^low cells expressed significant levels of GITR but not the other markers (not shown).

TGF-β1⁺ T cells from the test and control groups of mice were examined for the expression of various markers. As shown in Fig. 6, F and G, these T cells from the test group are primarily of the memory type (CD62L^low). Although the percentages of CD62L^lowTGF-β1⁺ cells ranged between 6.6 and 9.7, CD62L^highTGF-β1⁺ T cells were significantly lower (between 0.8 and 1.7%) in these mice (p = 0.0039). Examination of resting CD62L^low and CD62L^high subpopulations of TGF-β1⁺ T cells from the test group for Treg markers revealed that CD62L^low, but not CD62L^high, population expressed significant levels of GITR (Fig. 6,H). Although only a fraction of these cells was positive for Foxp3, both subpopulations demonstrated high levels of intracellular CTLA-4. IL-10 production and surface expression of CTLA-4 were not detectable in either population. In agreement with the above observations (Fig. 4), these results suggested that CD4⁺TGF-β1⁺ adaptive Tregs function through surface-bound TGF-β1 in a contact-dependent manner. Fig. 6 also shows that enhanced CTLA-4 engagement upon Ag presentation resulted in the generation of at least two distinct populations of Tregs. Examination of CD62L^low and CD62L^high subpopulations from the control group of mice showed that CD62L^high, but not CD62L^low, population expressed moderate levels of intracellular CTLA-4. Expression levels of GITR, IL-10, Foxp3, and surface CTLA-4 were not significant in either population.

CD62L^low and CD62L^high subpopulations of CD4⁺CD25⁺ and CD4⁺TGF-β1⁺ T cells from the test group of mice were sorted by high speed FACS (Fig. 7,A) and tested for their ability to suppress an Ag-specific effector T cell recall response. As shown in Fig. 7, B and C, TGF-β1⁺CD62L^low T cells suppressed effector T cell proliferation as effectively as CD25⁺CD62L^low Tregs (percentages of proliferated T cells were 8.5 and 10.4, respectively, compared with 32.7 in the presence of CD25⁻TGF-β1⁻ control T cells and 35.2 in the absence of Tregs). However, CD62L^high fractions of both CD25⁺ and TGF-β1⁺ T cells showed comparable efficacy (percentages of proliferated cells were 16.5 and 17.4, respectively), albeit lower compared with CD62L^low fractions, in suppressing the recall response. In addition, as evident from the MFI values, effector T cells cultured in the presence of CD62L^low adaptive Tregs failed to undergo multiple divisions compared with cells cultured in the presence of control T cells or naive Tregs. Although the adaptive Treg (CD62L^low) population demonstrated a significant level of suppression even at a suppressor:effector ratio of 0.125:1 (p < 0.021), relatively larger numbers of CD62L^high cells (at least 0.25:1 suppressor:effector ratio; p < 0.043) were needed to induce a significant level of suppression (Fig. 7,C). These observations demonstrated a higher suppressive efficacy of adaptive Tregs, presumably due to their specificity toward the cognate Ag. Although CD25⁺CD62L^low T cells from control mice failed to suppress the effector T cell response, the suppressive abilities of naive CD25⁺ T cells from control and test mice were comparable (not shown). This suggests that CD25⁺CD62L^low T cells from control mice, at least in part, are activated effector T cells and, therefore, do not suppress effector T cells efficiently. Significantly lower levels of various Tregs markers on these T cells from control mice compared with CD25⁺CD62L^low T cells from test mice as evident from Fig. 6 support this notion. Collectively, these observations suggest that the suppressive ability of CD4⁺CD25⁺ T cells from control mice observed in Fig. 4 is primarily mediated by the CD62L^high population.

One of the major characteristics of Tregs is their hyporesponsiveness to cognate Ag. Therefore, we tested whether the enriched population of Tregs from the test group of mice can proliferate upon stimulation with OVA-pulsed APCs or anti-CD3/CD28 Abs in the presence or absence of excess IL-2 (50 U/ml). Although stimulation using OVA failed to induce the proliferation of all four subpopulations, the addition of IL-2 induced the division of a considerable number (24.3 and 30.5% of cells with CFSE dilution) of memory (CD62L^low) Tregs (Fig. 7 D). Further, similar numbers of these memory Tregs underwent cell division(s) when stimulated with anti-CD3/CD28 Abs, which was further increased in the presence of exogenous IL-2 (35.6 and 58.0% of cells with CFSE dilution). The proliferative response of Tregs to strong anti-CD3/CD28 Ab stimulation alone and to cognate Ag in the presence of excess IL-2 revealed that strong activation signals can induce the expansion of adaptive Tregs.

To test the Ag specificity of DC-directed, CTLA-4 engagement-induced T cell hypoproliferation, an adoptive transfer experiment was conducted. HEL-primed mice were adoptively transferred with CFSE-labeled CD4⁺ T cells from OVA_(323–339) peptide-primed DO11.10 TCR-transgenic mice and treated with OVA_(323–339) peptide-pulsed control or anti-CTLA-4 Ab-coated DCs. Spleen and LN cells from recipient mice, obtained on day 4 posttransfer, were tested for CFSE dilution. The test group of mice showed significantly lower T cell proliferation compared with the controls (p < 0.0084) (Fig. 8,A). In a parallel experiment, unlabeled DO11.10 CD4⁺ T cell recipient mice were treated with Ab-coated DCs. Lymphocytes obtained on day 15 posttreatment were labeled with CFSE and further tested for T cell proliferation upon challenge with Ags ex vivo. CD4⁺ T cells from both groups responded similarly to HEL challenge ex vivo (Fig. 8,B, upper panel). However, the recall response of DO11.10 T cells from the test group of mice was significantly lower in terms of the number of proliferating cells (p = 0.0048) as well as the rate of proliferation (Fig. 8 B, lower panel). This demonstrated the Ag specificity of T cell hyporesponsiveness induced upon DC-directed enhanced CTLA-4 engagement.

To examine whether the Treg inducing potential of enhanced CTLA-4 engagement can be exploited to treat autoimmunity, CBA/J mice that were immunized with mTg to induce thyroiditis were injected with mTg-pulsed anti-CTLA-4 or control Ab-coated DCs and tested for immune response and disease outcome 15 days posttreatment. Significantly reduced lymphocyte infiltration into thyroids and minimal follicular destruction were noted in treated mice compared with controls (Fig. 9,A). Although 90% of the control mice developed grade 2–3 thyroiditis, only 30% of the mice from the test group showed grade 2 thyroiditis while the remaining 70% showed grade 1 thyroiditis. Importantly, as observed above using OVA as the candidate Ag (Fig. 6), mice from the test group showed a significant increase in Foxp3⁺ and TGF-β1⁺ adaptive Tregs compared with control mice (Fig. 9,B). Consistent with the results shown in Fig. 3, ex vivo challenge of spleen cells from Ab-coated DC-treated mice with mTg demonstrated that CD4⁺ T cells from these mice were not only hyporesponsive to mTg but could also produce significantly higher IL-10 and TGF-β1 and reduced IFN-γ and IL-4 compared with T cells from control mice (not shown). These observations suggest that adaptive Tregs also played a key role in the suppression of thyroiditis.

Next, we examined whether DC-directed CTLA-4 engagement confers a long-lasting therapeutic effect. Mice were immunized with mTg and treated with control or anti-CTLA-4 Ab-coated, mTg-pulsed DCs and rested for 60 days before rechallenging them with mTg. As shown in Fig. 10,A, whereas the control mice demonstrated grade 2 or 3 thyroiditis, the test group showed grade 0 or 1 thyroiditis 15 days postchallenge. Although T cell proliferative (Fig. 10,B) and adaptive Treg responses (not shown) remained more or less similar to that observed above in Fig. 9, the test group showed significantly reduced titers; mean IgG1 and IgG2a Ab titers were 8 × 10⁴ and 4 × 10⁴, respectively (p < 0.01) against mTg compared with 1.6 × 10⁵ and 2.4 × 10⁵ for the control group. Interestingly, challenge exposure to Ag did not appear to boost the Ab response. These observations led us to conclude that Ab response is also significantly suppressed, perhaps due to a decrease in the IFN-γ and IL-4 levels. The above results showed that disease suppression induced by DC-directed, enhanced CTLA-4 engagement is persistent even after rechallenge with the Ag and the inflammatory agent LPS.

Although a role for CTLA-4 in Treg function and TGF-β1 induced Treg generation has been suggested (14, 15, 16, 17, 18), understanding its direct role in Treg generation has remained elusive. Considering that CTLA-4 plays an important role in T cell negative regulation, we hypothesized that enhancing the CTLA-4 signaling strength at the immunological synapse upon Ag presentation may affect T cell differentiation. Therefore, we examined the ability of an agonistic anti-CLTA-4 Ab-coated Ag presenting matured DCs to suppress Ag-specific T cell response in vitro and in vivo. Our study demonstrated that the dominant engagement of CTLA-4 induced not only Ag-specific T cell hyporesponsiveness but, more importantly, a strong adaptive Treg response. In vivo studies showed that the ability of T cells from anti-CTLA-4 Ab-coated DC recipients to respond to challenge Ag exposure was dramatically reduced and that a majority of the responding T cells failed to undergo multiple cell divisions compared with T cells from control mice. Interestingly, anti-CTLA-4 Ab-coated DC treated mice showed a significant increase in the CD4⁺CD25⁺ T cell numbers and T cells from these mice produced appreciably higher amounts of suppressor cytokines such as IL-10 and TGF-β1 upon re-exposure to the Ag, suggesting their Ag specificity and potential mode of action.

Intriguingly, treatment with Ag-pulsed and anti-CTLA-4 Ab-coated DCs resulted in a significant increase in the number of memory (CD62L^low) CD4⁺CD25⁺ T cells. However, enhanced CTLA-4 engagement in vivo had no effect on the overall number of naive (CD62L^high) CD4⁺CD25⁺ T cells. Expression of Foxp3 by a large proportion of CD4⁺CD25⁺CD62L^low T cells from the test group compared with the control group provided further credence to their identity as Tregs. In agreement with this observation, T cells depleted of a CD4⁺CD25⁺ population showed an enhanced proliferative response and produced more inflammatory cytokines (e.g., IL-2 and IFN-γ) but showed reduced levels of the regulatory cytokines IL-10 and TGF-β1. Importantly, CD4⁺CD25⁺D62L^low Tregs more profoundly suppressed effector T cell response compared with CD4⁺CD25⁺D62L^high Tregs. Studies have shown that CD4⁺CD25⁺ T cells originate in the thymus after self-Ag-mediated selection and that these T cells can suppress self-Ag-specific T cells effectively in the periphery (24, 25, 26). Both thymic and peripheral naive CD4⁺CD25⁺ T cells express significant levels of CD62L on the surface (27, 28). Further, studies have also shown that the CD62L^high, but not the CD62L^low, population is more effective in suppressing effector T cell response (29, 30). Although the origin of CD4⁺CD25⁺D62L^low T cells is not well understood, they could be activated effector T cells, CD4⁺CD25⁺D62L^high Tregs that were activated through TCR ligation, and/or induced Tregs of CD4⁺CD25⁻ T cell origin. Therefore, despite the possible heterogeneity, the CD4⁺CD25⁺D62L^low T cell population with suppressor function was considered as adaptive Treg in this study.

Production of significantly higher amounts of IL-10 and TGF-β1 upon exposure to Ag by adaptive Tregs and reversal of their suppressive effects upon neutralization of these cytokines indicated that they are functionally activated upon cognate Ag binding and exert their function primarily through IL-10 and/or TGF-β1. Importantly, our results also showed that CD25⁺, but not the CD25⁻, CD4⁺ T cells from the test group secreted significant amounts of IL-10 and TGF-β1. Interestingly, we also noted that the CD4⁺ T cell population, depleted of CD25⁺ T cells, continued to suppress effector T cell response. This supported the notion that additional Treg population(s) may be present in the CD25⁻ fraction. Examination of CD25⁻CD4⁺ T cells showed a significant number of cells with surface-bound TGF-β1 in the test mice, and these cells exhibited a predominantly CD62L^low memory phenotype. This led us to conclude that this distinct subset of Tregs, at least in part, exerted its regulatory function in a contact-dependent manner through surface-bound TGF-β1 and not through secreted cytokines.

TGF-β1 can suppress effector T cells in both the free (secreted) and membrane-bound forms from T cells (31, 32, 33, 34). Although it is unknown how TGF-β1, which lacks a transmembrane domain, is expressed on the T cell surface, it is believed that it can bind to yet unidentified surface molecules (35, 36). Although we did not detect TGF-β1 on CD4⁺CD25⁺ T cell surface at the resting stage, CD4⁺CD25⁺CD62L^low T cells appeared to secrete significant amounts of TGF-β1. However, enriched T cells with surface-bound TGF-β1 did not secrete significant amounts of TGF-β1 or IL-10 even upon stimulation (not shown), suggesting that CD4⁺CD25⁺CD62L^low cells could be one of the possible sources of TGF-β1 found on CD4⁺CD25⁻ T cell surface.

It is not clear why adaptive Tregs are more efficient in suppressing Ag-specific effector T cell response compared with their natural counterparts. One possible explanation is that these Tregs themselves are Ag specific. Although hyporesponsive in terms of their proliferation, these Tregs can be activated upon antigenic stimulation to produce IL10 and TGF-β1, which suppress effector T cell function. The lack of specificity toward the foreign Ag may prevent natural Tregs from undergoing activation and thus render them less efficient in their suppressor activity compared with the adaptive Tregs. However, the required activation of natural Tregs (37, 38) to mediate the suppression of effector T cells may come from IL-2 produced by the effector T cells early after Ag exposure.

Although genetic approaches for expressing an anti-CTLA-4 single chain Ab for the selective engagement of CTLA-4 to suppress T cell response have been reported (39, 40), the importance of CTLA-4 signaling in the induction of adaptive Tregs has not been reported. Earlier, we showed that coating the target tissue or cells under immune attack with an anti-CTLA-4 Ab can result in an increase in the memory CD4⁺CD25⁺ T cell numbers (19, 20). Current results extended these earlier observations and further demonstrated that enhancing the strength of CTLA-4 signaling in T cells upon Ag presentation by DCs could induce at least two distinct populations of Tregs.

Although we do not know the time of engagement of CTLA-4 on T cells by the DC-bound anti-CTLA-4 Ab, we assume that it occurs predominantly upon T cell activation, which is required for CTLA-4 up-regulation. Similar levels of CD69 expression, an indicator of the early activation of T cells cultured in the presence of anti-CTLA-4 and control Ab-coated DCs (Fig. 2), indicated that CTLA4-mediated signaling may result from prolonged contact between T cells and DCs. We speculate that the increased strength of CTLA-4 engagement in the immune synapse, concurrently or immediately after the Ag-specific activation of T cells, not only can suppress their proliferation but can also cause increased IL-10 and TGF-β1 and significantly reduced IL-2 and IFN-γ, production by T cells (Fig. 2), thus creating a microenvironment conducive for Treg generation. Other studies have shown that IL-10 and TGF-β1, individually or in combination, can act in an autocrine and/or a paracrine manner and promote adaptive Treg generation under different conditions (41, 42). It is important to note that engagement of CTLA-4 by B7.1 and B7.2 present on the DCs alone may not be sufficient to induce IL-10 and TGF-β1 responses and Tregs. However, controlling T cell activation by increasing the CTLA-4 specific ligand strength on APCs as demonstrated here can skew the cytokine response from the inflammatory to the suppressive type, which, in turn, can influence Treg induction.

If these adaptive Tregs are truly Ag specific, we wondered whether CTLA4 can be selectively engaged to treat autoimmunity. Because the Ag-specific TCR and the enhanced T cell repressor signal are concomitantly delivered by the DCs, the resulting hyporesponsiveness most likely will be Ag specific. In this context, when OVA peptide-specific T cells were introduced to HEL-primed mice followed by treatment with OVA peptide-pulsed, anti-CTLA-4 Ab-coated DCs (Fig. 8), only OVA peptide-specific, but not HEL-specific T cells, showed significant hypoproliferative response to the Ag. More importantly, the ability of mTg-pulsed anti-CTLA-4 Ab-coated DCs to suppress thyroiditis, characterized by a profound hypoproliferative response of T cells against mTg, showed that dominant engagement of CTLA-4 upon Ag presentation is an effective way to suppress autoimmunity. In addition, enhanced CTLA-4 engagement by anti-CTLA-4 Ab-coated DCs could produce long-lasting disease suppression, and rechallenge with mTg failed to overcome the treatment-induced T cell hyporesponsiveness. Intriguingly, dominant CTLA-4 engagement affected both Th1 and Th2 T cells as evident from the cytokine and subclass Ab levels.

Collectively, we have not only shown that selective CTLA-4 signaling can play an important role in adaptive Treg induction, but also that the DC-directed, selective CTLA-4 engagement approach is an effective way to treat autoimmunity. This approach exploits the unique Ag-presenting properties of DCs to harness the potential of CTLA-4-mediated active T cell inhibition to induce Ag-specific Tregs. This approach is highly flexible and can be readily adapted to a wide range of Ags for treating autoimmunity and transplant rejection.

The authors have no financial conflict of interest.