Regulatory T cells inhibit acute IFN-γ synthesis without blocking T-helper cell type 1 (Th1) differentiation via a compartmentalized requirement for IL-10

CD28 Uncouples Suppression of Proliferation and Effector Cytokines.

The study of Treg effects on differentiation has been hindered by their potent inhibition of T-cell proliferation. CD28 signaling in target T cells can override Treg suppression of T-cell proliferation (15–17) and may facilitate the study of proliferation-independent Treg effects. To test this notion, we measured cytokine secretion by CD4 target T cells early during Th1 priming with anti-CD3 and anti-CD28 mAbs in the presence Tregs. As previously reported (15), CD28 blocks Treg suppression of target T-cell IL-2 early before division (10 h; Fig. 1 A and B) enabling subsequent division (36 h; Fig. 1A). At 36 h, however, target T cells secreted IFN-γ when cultured alone but failed to secrete IFN-γ when cultured with Tregs (Fig. 1 A and B) despite robust proliferation (Fig. 1A) over a range of Treg:target T cell ratios (Fig. 1C). Tregs led to a decrease in IFN-γ frequency and amount per cell (Fig. 1B). Addition of nonregulatory T cells, to control for cell density, did not reduce IFN-γ (Fig. 1 A and B). Hence, CD4⁺ T cells remain susceptible to Treg suppression of IFN-γ despite provision of CD28 signals (Fig. 1A) with a selective loss in the number of IFN-γ producers (Fig. 1D). Thus, regulation of IFN-γ production is independent of regulation of proliferation.

Early Treg Blockade of IFN-γ Is Independent of Cytokine Consumption or Production of Suppressive Cytokines.

To determine when and how Tregs suppress IFN-γ, we looked at the kinetics of IFN-γ production. Secretion of IFN-γ by naïve target T cells was observed as early as 6–12 h poststimulation and peaked at 36 h. With Tregs, IFN-γ secretion was induced between 6–12 h but was not enhanced 18–36 h into the culture (Fig. 1E), even at high concentrations of the polarizing cytokine recombinant IL-12 (rIL-12) (Fig. S1B). Tregs could passively limit positive signals for Th1 differentiation (18) or actively secrete suppressive cytokines. Th1 differentiation depends, in addition to IL-12, on IL-2 and IFN-γ signaling (19, 20). However, neither cytokine appeared limiting in Treg cocultures, because cytokine receptor signaling in target T cells was intact (Fig. S1A), and exogenous provision of IL-2 or IFN-γ could not rescue IFN-γ production (Fig. S1B). Tregs also may regulate immunity through consumption of key cytokines; however, Tregs derived from IFN-γ receptor(IFN-γR)– and IL-12 receptor (IL-12R)–deficient mice maintained inhibitory function in Th1 cultures (Fig. 1F). Loss of IFN-γ was also not associated with increased cell death of target T-cells by 7-amino-actinomycin D (7AAD) or expression of antiapoptotic B-cell lymphoma 2 (Bcl-2) (Fig. 1G). Therefore, we conclude that Tregs do not suppress IFN-γ secretion by interrupting Th1 differentiation by limiting key cytokines.

We next tested the requirement for immunosuppressive cytokines (21–25) in suppressing IFN-γ production. Target T cells isolated from dominant-negative TGF-β receptor type II (TGF-βRII^−/−) mice (which are unable to respond to TGF-β) remained sensitive to Treg inhibition of IFN-γ (Fig. S2A), suggesting that suppression of IFN-γ is independent of direct TGF-β signaling in target T cells. Similarly, suppression of IFN-γ remained intact in the absence of IL-35, using Epstein-Barr virus-induced gene 3 (Ebi3^−/−) or p35^−/− Tregs (Fig. S2B) and IL-10 (Fig. S2 C and D). Therefore, suppressive cytokines are not essential for acute regulation of IFN-γ by Tregs during Th1 priming. Consistent with IFN-γ suppression being independent of cytokine consumption or a Treg-secreted cytokine, transwell cultures in which Tregs were separated physically from the target T cells failed to support IFN-γ inhibition (Fig. 1H). Thus, Tregs require close proximity with the targets (and/or antigen-presenting cells, APCs) to mediate IFN-γ suppression during Th1 priming.

Acute Regulation of IFN-γ but No Inhibition of Lineage Commitment.

Thus far we have observed Treg inhibition of IFN-γ production during the early stages of Th1 differentiation. To determine if Tregs disrupt T-helper cell differentiation itself, target T cells were isolated after primary Th1 culture with Tregs and were recultured for 6 d without Tregs and without polarizing cytokines. Th1-primed target T cells expanded and produced IFN-γ but not IL-4 (Fig. 2 A and B) (or IL-17) regardless of primary culture with Tregs. To test functional flexibility, Th1-primed target T cells cultured with Tregs were recultured in Th2-polarizing conditions in the absence of Tregs. Compared with Th2 cells, Th1-primed target T-cell populations primed with or without Tregs made significantly less IL-4 (Fig. 2C), demonstrating a functional commitment to the Th1 lineage. Our data suggest Tregs acutely inhibit IFN-γ but do not inhibit commitment to the Th1 differentiation program. An extension of this idea would predict that Tregs also should be capable of acute regulation of IFN-γ production from established Th1 effectors. Indeed, once differentiated, Th1 cells remain susceptible to acute regulation of IFN-γ by Tregs (Fig. 2D), as previously proposed (26).

Regulation of IFN-γ Transcripts Independent of Tbet.

To determine how Tregs modulate Th1 function, we performed a targeted transcriptional array for genes associated with CD4⁺ T-cell differentiation. Target T-cell mRNA was isolated from Th1-primed cultures in the presence or absence of Tregs. The presence of Tregs led to the selective transcriptional down-regulation of IFN-γ at 36 h (Fig. 3A) with kinetics matching that seen for IFN-γ protein (Fig. 1E and Fig. S3A). However, many genes, including other Th1 effector molecules such as TNF-α and the Th1 lineage-specific transcription factor Tbet, were not modulated by Tregs at 36 h (Fig. 3A). Consistent with effective Th1 differentiation, GATA-binding protein 3 (GATA3) was not induced in the presence or absence of Tregs (Fig. S3B).

To assess potential posttranscriptional control of Tbet expression, we measured Tbet protein levels by FACS. Similar levels of Tbet expression were found in target T cells cultured alone or with Tregs throughout the first 36 h (Fig. 3B and Fig. S3C), despite ongoing Treg suppression of IFN-γ (Fig. 3B). Tbet appears to be expressed in two independent waves with a late (post-48 h) IL-12/signal transducer and activator of transcription 4 (STAT4)-dependent stage being critical for Th1 lineage commitment and high-level IFN-γ production (27). To determine if Tregs interfere with this late Th1 lineage decision, we examined STAT4 signaling and sustained Tbet expression. STAT4 signaling was largely unaffected by Tregs in the late phase (48 h and beyond) (Fig. 3C), consistent with prolonged Tbet expression (Fig. 3D). As a measure of Tbet function, we assessed expression of the Tbet gene target chemokine (C-X-C motif) receptor 3 (CXCR3) and found no difference between target T cells cultured alone and with Tregs (Fig. 3E). Thus, Tregs reduce transcript levels of IFN-γ without blocking Tbet expression. Intact Tbet likely accounts for our observed imprinting of Th1 function in target T cells primed with Tregs (Fig. 2).

Control of Effector Function Independent of Differentiation in Vivo.

To test Treg modulation of Th1 differentiation formally in vivo, we established a model to identify the initial acquisition of CD4 effector function directly. Cytokines were measured directly ex vivo by intracellular cytokine staining in the presence of Brefeldin A, eliminating the need to reactivate cells in vitro, similar to the methods of McLachlan et al. (26). Using DO11.10⁺ T-cell receptor (TCR) transgenic mice expressing a Foxp3-GFP reporter, we isolated naïve DO11.10⁺ target T cells (CD4⁺CD62L^hiCD44^loGFP⁻Thy1.2⁺) and DO11.10⁺ Tregs (CD4⁺GFP⁺Thy1.2⁺). Carboxyfluorescein succinimidyl ester (CFSE)-labeled naïve T cells were transferred with or without Tregs to immunocompetent Thy1.1⁺ WT BALB/c mice immunized with ovalbumin peptide in incomplete Freund's adjuvant (OVAp/IFA). On days 3 and 5 after immunization, transferred target T cells (Thy1.2⁺ Foxp3⁻) were assessed for proliferation, expansion, and cytokines.

As in previous reports, Tregs did not prevent target T-cell entry into division (Fig. 4A) (14) but did compromise expansion (total number of target T cells: 2.1 × 10⁶ ± 0.2 target T cells alone vs. 0.5 × 10⁶ ± 0.6 target T cells with Tregs) (see Fig. 4B for IFN-γ numbers) (13). In addition to a reduced number of primed effector cells, the remaining activated pool was compromised in its ability to make IFN-γ. Direct ex vivo IFN-γ production was measured early in the draining lymph node (on days 3 and 5) and in tissue at the site of immunization (ear dermis) (on day 5; no transferred cells were detected day 3). Target T cells produced IFN-γ at both sites, but this production was suppressed by twofold in the presence of Tregs in both the lymph node (Fig. 4B) and inflamed skin (Fig. 4 C and D). Similar results were obtained following OVAprotein/IFA immunization (Fig. S4). Consistent with our in vitro results, inhibition of IFN-γ was independent of Tbet expression in both the lymph node and ear (Fig. 4E). In vitro Treg regulation was specific for IFN-γ but not for TNF-α. Although we were unable to detect TNF-α using the direct ex vivo approach, a brief ex vivo restimulation revealed a specific loss of IFN-γ⁺TNF-α⁺ dual producers independent of TNF-α single producers (Fig. 4F). Thus, Tregs regulate at least two distinct components of immunity in vivo: the expansion of an effector pool (four- to fivefold) and the regulation of IFN-γ production (twofold) in a Tbet-independent fashion.

Our data suggest Tregs do not limit acquisition of CD4 effector potential but acutely shutdown execution of function. To test this notion formally, we isolated day 5 in vivo-primed DO11.10⁺ target T cells away from the transferred DO11.10⁺ Tregs and “parked” them in WT mice. The recipient mice were challenged with OVAp/IFA 5 d after cell transfer, and IFN-γ was measured directly ex vivo on day 3 (Fig. 4 G–J; labels in parentheses indicate conditions of initial priming). The DO11.10⁺ target T cells cycled (Fig. 4G) and made similar amounts of IFN-γ whether initially primed in the presence or absence of Tregs (Fig. 4 H and J). Consistent with a Th1 phenotype, they expressed Tbet but not GATA3 (Fig. 4 H and I). Therefore, DO11.10 T cells primed in the presence of Tregs retain the ability to mount a subsequent Th1 response when activated in the absence of Tregs. We conclude that Tregs do not impact the Th1 differentiation process directly; rather, they limit the pool of IFN-γ producers by suppressing the expansion of primed cells and also by acute regulation of IFN-γ production in the lymph node and tissue.

Compartmentalized Requirement for IL-10 by Tregs in Vivo.

To address mechanisms of IFN-γ regulation, we tested the role of Treg IL-10 in vivo using Tregs from DO11.10 Foxp3/GFP IL-10–deficient mice. In line with our results in vitro, IL-10–deficient Tregs retained the ability to suppress both expansion and IFN-γ production during the early stages of Th1 priming in the lymph node (Fig. 5A). Nonetheless, despite trafficking to the OVAp/IFA-inflamed dermis, similarly to WT Tregs (Fig. 5B), WT Tregs reduced the frequency of IFN-γ–producing CD4 T cells by twofold in the ear (P = 0.0027, frequency of IFN-γ producers in absence vs. presence of WT Tregs; n = 16) but the IL-10–deficient Tregs were unable to regulate IFN-γ production in the tissue (P = not significant, frequency of IFN-γ producers in absence vs. presence of IL-10^−/− Tregs; n = 10) (Fig. 5 C and D). These results (Fig. 5 A and D) demonstrate that Tregs use different regulatory mechanisms to limit IFN-γ production in the lymph node and in the inflamed tissue. That Treg-derived IL-10 is essential for IFN-γ regulation in the inflamed dermis is consistent with the selective accumulation of IL-10–producing Tregs in areas of tissue inflammation (28) and the breakdown in control at barrier surfaces when IL-10 is specifically deleted in Foxp3⁺ cells (11).

Mice.

WT BALB/c, C57BL/6, Thy1.1, DO11.10⁺, Foxp3GFP BALB/c, DO11.10⁺Foxp3GFP, dnTGF-βRII^−/−, and IL-10^−/− mice were from Jackson Laboratories. IL-10^−/− DO11.10⁺Foxp3GFP mice were bred and maintained at the University of Rochester (Rochester, NY). IL-12Rβ1^−/− and IL-12Rβ2^−/− mice were provided by D. Fairweather (The Johns Hopkins University, Baltimore, MD) and E. M. Lord (University of Rochester, Rochester, NY), respectively, and Ebi3^−/− mice were obtained from R. S. Blumberg (Harvard University, Cambridge, MA).

Cell Purification.

CD4⁺CD44^low naïve CD4⁺ T cells and CD4⁺ CD25⁺ Tregs were isolated using standard MACS columns (Miltenyi Biotech) or sorting by FACS (BD Bioscience). Further details can be found in SI Materials and Methods.

In Vitro Suppression Assay.

Target T cells (10⁷/mL) were labeled with 1 μM CFSE (Molecular Probes) for 5 min at room temperature. Then 1 × 10⁵ CD25⁻ Thy1.1⁺ CD4⁺ T cells were cultured with 1 × 10⁵ APC, 1 μg/mL anti-CD3 mAb (2C11), and 1 μg/mL anti-CD28 mAb (37.51) with or without 1 × 10⁵ CD25⁺ Thy1.2⁺ CD4⁺ T cells in 96-well U-bottomed plates. Th1-priming conditions were 10–20 ng rIL-12, anti-IL-4 Ab 11B11, 10 units/mL of recombinant human IL-2 (rhIL-2). Cells were cultured in complete RPMI 1640 with 10% (vol/vol) heat-inactivated FCS (Hyclone).

Quantitative Real-Time RT-PCR.

RT-PCR was performed using standard techniques. Further details can be found in SI Materials and Methods.

Flow Cytometry.

All FACS analyses used standard intracellular staining techniques or the cytokine capture assay, as described (15). Further details can be found in SI Materials and Methods.

In Vivo Suppression Assay.

For the in vivo suppression assay, 0.5–1 × 10⁶ sorted naive CD4⁺CD62L^highCD44^low KJ-126⁺ cells were CFSE labeled (5 μM) and injected i.v. alone or with equal numbers of sorted CD4⁺Foxp3GFP⁺KJ-126⁺ Tregs into congenic Thy1.1 BALB/c mice. Mice were immunized intradermally in the ear with OVAp/IFA (1 μM OVAp per ear). Three to five days after immunization cells were isolated from the ears and draining lymph node in buffers containing Brefeldin A and were assessed ex vivo for function. Where indicated, on day 5 after immunization target T cells (Thy1.2⁺ Foxp3GFP⁻) were sorted, and 100,000 target T cells were injected alone into fresh congenic mice and rested for 5 d before challenge with OVAp/IFA. Further details can be found in SI Materials and Methods.

Statistical Analysis.

A paired student's t test was performed for the in vitro data. A Mann–Whitney test was preformed for the in vivo data. *P ≤ 0.05, **P ≤ 0.005, and ***P ≤ 0.0005.