Cutting Edge: TGF-β1 and IL-15 Induce FOXP3⁺ γδ Regulatory T Cells in the Presence of Antigen Stimulation¹

Human γδ T cells are a lymphocyte population with unusual tissue distribution and Ag recognition pathways. Vγ9Vδ2 T cells are a major subset of the circulating γδ T cell pool and are involved in immunity against pathogens and malignancies (1, 2, 3, 4, 5, 6, 7, 8, 9). Unlike classical αβ T cells, Vγ9Vδ2 T cells recognize nonprocessed Ags such as pyrophosphomonoesters (10). This recognition is mediated by the TCR and is not restricted by MHC molecules (11, 12). A regulatory role for γδ T cells has already been shown in a number of studies and is demonstrated by their influence on dendritic cells (13), neutrophils (14), and B cells (15) in humans, and on neutrophils (16) and macrophages (17, 18) in mice.

The immune system is responsible for protective response against pathogens and tumors and maintains immune tolerance. Regulatory T cells (Tregs)³ identified as Forkhead/winged helix transcription factor box P3 (FOXP3)-expressing CD4⁺CD25^high T cells represent a critical aspect in the balance of the immune response, preventing the development of autoimmune diseases (19) and preserving the efficiency of antitumor immune responses (20). To date, FOXP3 is the master gene that controls Treg development/function and is currently used as the major Treg marker (21, 22). In addition, IL-10-producing T regulatory type 1 (Tr1) cells and TGF-β-producing Th3 cells induced from uncommitted peripheral CD4⁺ T cells possess suppressive functions similar to those of naturally occurring Tregs (nTregs) (23, 24). The potential regulatory role of γδ T cells has been recently suggested but not yet clarified (17, 25, 26). In this study, we show for the first time that a subset of FOXP3-expressing Vδ2 T cells (Vδ2 Tregs) is induced under opportune Ag stimulation and cytokines.

PBMC from healthy donors were isolated from fresh buffy coats by Lympholyte (Cedarlane) density gradient centrifugation. Cells were cultured at a concentration of 4 × 10⁶/ml in the presence of IL-2 (6.5 U/ml), IL-15 (10 ng/ml) (both from Sigma-Aldrich), TGF-β1 (1.7 ng/ml; Calbiochem) and isopentenyl pyrophosphate (IPP; 20 μg/ml; Sigma). On days 3, 6, and 9, half of the supernatant volume was discarded and replaced with fresh medium containing cytokines. Aliquots of PBMC and bead-sorted γδ T cells (Miltenyi Biotec) were used fresh or were frozen for use at later time points.

Cells were surface stained with anti-Vδ2, anti-CD25, anti-CD69, anti-CD45RA, anti-CD45RO, anti-CD27, and anti-CD127 (BD Biosciences), fixed and permeabilized with saponin or Perm/Fix solution, and finally stained intracellularly with anti-CTLA-4 (BD Biosciences) or FOXP3 (Biolegend) according to the manufacturer’s instructions. The cells were analyzed using a FACScalibur instrument with CellQuest software (BD Biosciences). Because FOXP3 is expressed intracellularly, it is not possible to use it as a marker of the Treg phenotype for purification by sorting Vδ2⁺FOXP3⁺ cells. Therefore, Vδ2⁺ cells were purified from PBMC cultured to the point where the Vδ2⁺FOXP3⁺ population represented 30% or more of the Vδ2⁺ cells (35.6 ± 9.7; range, 30.1–50.4), using a FACSVantage-SE cell sorter (95% purity; BD Biosciences).

The suppressive effect of the FOXP3⁺-rich Vδ2 T cells on autologous PBMC was analyzed. First, PBMC were stained with carboxyfluorescein diacetate succinimidyl ester (CFDA-SE; 6 μg/ml; Molecular Probes and Invitrogen) in PBS-FCS (5%) and stimulated with plate-bound anti-CD3 (5 μg/ml; BD Pharmingen) and anti-CD28 (10 μg/ml; eBioscience) Abs. The flow-sorted FOXP3⁺-rich Vδ2 T cells were then added at different ratios, and the mixed lymphocyte cultures were incubated for 5 days at 37°C in a 5% CO₂-containing atmosphere. Bead-sorted γδ T cells were used as FOXP3-negative controls. IL-2 (6.5 U/ml; Sigma-Aldrich) was present in all cultures. Some cultures contained an anti-TGF-β1-blocking Ab (10 μg/ml; R&D Systems) or indomethacin (1 μg/ml; Sigma-Aldrich). In some experiments, PBMC and “effector cells” were separated by a semipermeable membrane (0.4-μm pore size, Falcon; Costar). Cell proliferation was measured by CFDA-SE through decreasing relative fluorescence intensity by half during each round of cell division by flow cytometry and analyzed by Flow Jo software (Tree Star).

Sorted FOXP3⁺-rich Vδ2 T cells were stimulated with IPP for 18 h and then supernatants were collected. The cytokine measurements were performed using a capture ELISA (R&D Systems) or cytometric bead assays for Th1/Th2 cytokines (BD Biosciences).

Nonparametric Mann-Whitney U test and Wilcoxon matched pairs test were used to evaluate the statistical significance. Differences were considered significant at p < 0.05.

In the presence of TGF-β1, CD4⁺CD25⁻ lymphocytes may differentiate into FOXP3⁺ Tregs, suppressing T cell proliferation (27). As a first step, we analyzed the expression of FOXP3 in Vδ2 T cells. Fig. 1 shows that the frequency of FOXP3⁺ Vδ2 T cells in freshly isolated PBMC is very low (1.6 ± 0.7%). Because Vδ2 T cells can be selectively stimulated using phosphoantigen and IL-2 (10), we analyzed the expression of FOXP3 on Vδ2 T cells under IPP/IL-2 stimulation in the presence of Treg-promoting cytokines such as IL-15 and TGF-β1 (28, 29). After 4 days of culture, IL-2/IPP-stimulated Vδ2 T cells expressed higher levels of FOXP3 in the presence of TGF-β1 and IL-15 (35.6 ± 9.7%; range, 30.1–50.4%) than in their absence (7.9 ± 4.9%) (Fig. 1, A and B). Interestingly, IL-15 or TGF-β1 alone (Fig. 1,C) may induce a considerable expression of FOXP3 on Vδ2 T cells, but lower than that obtained in combination. Experiments using only cytokines without IPP were also performed, showing that FOXP3 induction requires Ag stimulation (data not shown). Furthermore, FOXP3⁺ Vδ2 T cells were able to proliferate as well as FOXP3⁻ Vδ2 T cells (Fig. 1 D). These data suggest that the conventional Treg cytokine milieu may induce the expression of FOXP3 in Vδ2 T cells.

FOXP3 may be transiently expressed in activated nonregulatory T cell populations (30, 31). To assess the transient-vs-stable expression of the FOXP3 in Vδ2 T cells, we performed FOXP3 staining daily during a 10-day period. Fig. 2 A illustrates the kinetics of FOXP3 expression, which started to increase after 1 day of culture (day 1, 3.6 ± 0.9 vs day 0, 1.6 ± 0.7; p < 0.05), peaked on day 3 (day 3, 38.5 ± 10.7 vs day 1, 3.6 ± 0.9; p < 0.05), and remained high over 10 days.

Moreover, FOXP3⁺ Vδ2 T cells expressed significantly higher levels of CD45RA (17.9 ± 5.3 vs 6.45 ± 2.4; p < 0.05) and CD69 (16.1 ± 6.7 vs 6.3 ± 1.5; p < 0.05) than FOXP3⁻ Vδ2 T cells. Higher levels of CTLA-4 (215 ± 77.2 vs 113 ± 61.3), CD25 (113.6 ± 54.8 vs 74.4 ± 39.3), and HLA-DR (36.5 ± 6.3 vs 23.7 ± 11.6) were also observed, although the median fluorescence intensity values were not significantly different. The expression of CD27 (13.5 ± 8.0 vs 12 ± 7.9) and CD45R0 (20.2 ± 10.5 vs 15.9 ± 5.1) was similar in both populations. CD127, which has been reported to be inversely correlated with FOXP3 expression in human CD4⁺ Treg cells (32), was comparably expressed by FOXP3⁺ (11.9 ± 3.4) and FOXP3⁻ (15 ± 4) Vγ9Vδ2 T cells (Fig. 2 B). Thus, FOXP3⁺ Vδ2 T cells display an activated phenotype with an increased expression of CTLA-4 molecules that are also present on the cell surface of both adaptive and natural αβ Tregs (33).

We then evaluated the functional ability of cultured Vδ2⁺ cells, of which a high proportion (>30%) express FOXP3, to inhibit PBMC proliferation. The stimulation of CFDA SE-labeled PBMC with anti-CD3/anti-CD28 Ab-induced cell proliferation (Fig. 3,A) was significantly reduced in the presence of flow-sorted, FOXP3⁺-rich Vδ2 T cells (Fig. 3,C). The addition of ex vivo bead-purified Vδ2⁺ γδ T cells, which are almost entirely FOXP3-negative, had no inhibitory effect (Fig. 3,B). Surprisingly, the TGF-β1-neutralizing Ab was not able to totally abrogate the inhibitory activity of Vδ2 Tregs (Fig. 3,D), whereas FOXP3⁺-rich Vδ2 T cells produced high amounts of TGF-β1 (2,231.2 ± 390.2 pg/ml) (Fig. 3,G). Furthermore, FOXP3⁺-rich Vδ2 T cells produced moderate amounts of IL-4 (108.3 ± 39.1 pg/ml) and low levels of IL-6 (7.6 ± 7.4 pg/ml) and IL-10 (1.8 ± 2.1 pg/ml) (Fig. 3,G). Given that, to assess the role of cell contact in the inhibitory activity we performed Transwell experiments, which showed that PBMC proliferation was restored when FOXP3⁺-rich Vδ2 T cells were physically separated from effector PBMC (Fig. 3, E and F). These results suggest that, differently than T regulatory type 1 (Tr1) and Th3 regulatory cells, which suppress largely by cytokines, the feature of FOXP3⁺-rich Vδ2 T cells appear to be similar to that of CD4⁺CD25^high naturally occurring Tregs (nTregs) (34, 35). Moreover, the level of the proliferation-blocking effect was dependent on the ratio of PBMC to FOXP3⁺-rich Vδ2 T cells; the strongest effect was observed at a 1:1 target:FOXP3⁺-rich Vδ2 T cell effector ratio (Fig. 3 H).

We conclude that the TGF-β1 signaling pathway, known to regulate many immune processes, appears to be also involved in inducing γδ Tregs similarly as it does for αβ Tregs. Differently from αβ, the combination of TGF-β1, IL-15, and TCR stimulation may be central in this process, whose detailed molecular mechanisms nevertheless remain to be clarified. Vδ2 Tregs may be expanded and potentially exploited therapeutically in a variety of clinical situations.

We dedicate this work in memoriam to our dearest friend Fabrizio Poccia, who was ne plus ultra of Vγ9Vδ2 T cell research. It was a great privilege and an immense pleasure for us to have been associated with Fabrizio. We will miss forever his friendship and infectious enthusiasm for immunological investigations.

The authors have no financial conflict of interest.