Identification of T helper type 1–like, Foxp3⁺ regulatory T cells in human autoimmune disease

Abstract

CD4⁺CD25^highCD127^low/− forkhead box p3 (Foxp3)⁺ regulatory T cells (T_reg cells) possess functional plasticity. Here we describe a higher frequency of T helper type 1 (T_H1)-like, interferon-γ (IFN-γ)-secreting Foxp3⁺ T cells in untreated subjects with relapsing remitting multiple sclerosis (RRMS) as compared to healthy control individuals. In subjects treated with IFN-β, the frequency of IFN-γ⁺Foxp3⁺ T cells is similar to that in healthy control subjects. In vitro, human T_reg cells from healthy subjects acquire a T_H1-like phenotype when cultured in the presence of interleukin-12 (IL-12). T_H1-like T_reg cells show reduced suppressive activity in vitro, which can partially be reversed by IFN-γ–specific antibodies or by removal of IL-12.

Multiple sclerosis is a genetically mediated chronic inflammatory disease of the central nervous system (CNS). Activated CD4⁺ inflammatory cells in the circulation of affected individuals infiltrate into the CNS and damage both myelin and axons^1,2. A general loss of immune regulation is commonly seen in human autoimmune diseases^3–6.

In light of recent findings suggesting that Foxp3⁺ T_reg cells show functional and phenotypic plasticity and are capable of secreting proinflammatory cytokines^7–9, we were interested in analyzing ex vivo secretion of cytokines by T_reg cells from individuals with relapsing/remitting multiple sclerosis (RRMS) that had not been receiving any immunomodulatory treatment as compared to age-matched healthy controls (Supplementary Methods and Supplementary Table 1). We stimulated FACS-sorted CD4⁺CD45RA⁻CD25^highCD127^low/− T_reg cells (Supplementary Fig. 1a) for 4 h with phorbol 12-myristate 13-acetate (PMA) and ionomycin. The percentage of T_reg cells producing IFN-γ was significantly higher in untreated individuals with RRMS as compared to healthy control individuals (Fig. 1a,b). The frequency of T_reg cells producing IL-17 did not differ between individuals with RRMS and control subjects. Analysis of the Foxp3⁺ T_reg cell–specific demethylated region (TSDR) in sorted IFN-γ⁺Foxp3⁺ and IFN-γ⁻Foxp3⁺ multiple sclerosis T_reg cells revealed that IFN-γ–producing Foxp3⁺ T_reg cells possessed a similar pattern of demethylation in the TSDR region to that of IFN-γ⁻Foxp3⁺ T_reg cells (Fig. 1c) and healthy control Foxp3⁺ T_reg cells (data not shown).

Figure 1.

T_reg cells from individuals with RRMS secrete IFN-γ ex vivo. (a) The frequency of FACS-sorted IFN-γ⁺ and IL-17⁺ T_reg cells in healthy control individuals (left) and untreated individuals with RRMS (middle, n = 17) gated on Foxp3⁺ T_reg cells. Right, purity analysis of the sorted IFN-γ⁺Foxp3⁺ and IFN-γ Foxp3⁺ populations from subjects with RRMS used for methylation analysis in c. (b) Percentage of IFN-γ⁺Foxp3⁺ and IL-17⁺Foxp3⁺ T_reg cells (n = 17) as a proportion of total Foxp3⁺ T_reg cells. (c) Representative example of methylation analysis of the TSDR region of the FOXP3 locus in sorted IFN-γ⁺Foxp3⁺ and IFN-γ⁻ Foxp3⁺ T_reg cells from subjects with RRMS. An analysis of IFN-γ⁺Foxp3⁻ memory T cells from subjects with RRMS is shown as a control. (d) Proliferation of responder T (T_resp) cells cultured with ex vivo FACS-sorted T_reg cells from healthy control subjects and untreated subjects with multiple sclerosis (MS; T_reg cell:T_resp cell ratio of 1:2) in the presence or absence of an IFN-γ–specific antibody (n = 4). (e) The frequency of IFN-γ⁺ and IL-17⁺ T_reg cells in healthy control subjects (left) or IFN-β–treated patients with RRMS (right) as assessed by intracellular cytokine staining and FACS analysis. The bar diagram (right) shows the percentage of IFN-γ⁺Foxp3⁺ and IL-17⁺Foxp3⁺ cells as a proportion of total Foxp3⁺ T_reg cells in healthy controls or IFN-β–treated patients with RRMS (n = 12). Approval for studies was obtained from the Brigham and Women’s Hospital Institutional Review Board, and informed consent was obtained from all donors.

After a 4-h stimulation with PMA and ionomycin, T_reg cells isolated ex vivo from untreated subjects with RRMS showed a T_H1-like phenotype, including secretion of IFN-γ, upregulation of mRNA expression of the T_H1-associated transcription factor TBET (encoded by TBX21) and downregulation of RORC (encoding RAR-related orphan receptor C) and TGFB1 (encoding transforming growth factor β1) mRNA (Supplementary Fig. 2 and Supplementary Table 1). CXCR3, but not CCR5 or IL10, was upregulated in T_reg cells from subjects with RRMS as compared to healthy controls (Supplementary Fig. 2). T_reg cells from subjects with RRMS downregulated CTLA4 (encoding cytotoxic T lymphocyte–associated protein 4) (271.8 ± 86.9 (arbitary units) in controls compared to 43.91 ± 11.7 in RRMS T_reg cells).

To ascertain whether IFN-γ secretion by RRMS T_reg cells reduces their suppressive activity, we cultured T_reg cells and responder T cells ex vivo in the presence of an IFN-γ–specific antibody. The suppressive activity of multiple sclerosis T_reg cells was significantly increased upon IFN-γ blockade, whereas healthy control T_reg cells were not affected (Fig. 1d).

IFN-β has been shown to affect the IL-12–IFN-γ axis in people with multiple sclerosis¹⁰, among its other immunomodulatory effects. To examine the possible in vivo role of this axis in the generation of IFNγ⁺ Foxp3⁺ T cells in people with RRMS, we performed a cross-sectional investigation, isolating T_reg cells from patients with RRMS treated for at least 1 year with IFN-β and comparing them to T_reg cells from healthy control subjects. Intracellular staining revealed that the frequency of IFN-γ⁺Foxp3⁺ T cells in IFN-β–treated subjects with RRMS was similar to that in age- and sex-matched healthy control subjects (Fig. 1e). This was accompanied by an increase in the frequency of IL-17⁺Foxp3⁺ T cells in IFN-β–treated patients with RRMS compared to healthy controls (Fig. 1e). Although these alterations in cytokine release could be due to a direct effect of IFN-β on IL-17 secretion by T_reg cells, IFNβ–mediated decreases in the amount of IL-12 could also induce a change in the cytokine milieu, driving increased IL-17 production.

Individuals with autoimmune disease have elevated IL-12 expression, and T_reg cells isolated from such individuals show reduced suppressive activity^3,11,12. To elucidate the mechanism by which T_reg cells produce IFN-γ, we hypothesized that IL-12, a cytokine associated with T_H1 responses, would induce a T_H1-like phenotype in T_reg cells, similar to the phenotype we observed ex vivo in T_reg cells from subjects with RRMS. We cultured T_reg cells from healthy control subjects in the presence or absence of IL-12. After 4 d, a significant (P = 0.009) percentage of IL-12–stimulated T_reg cells secreted IFN-γ (Fig. 2a and Supplementary Fig. 1b). This increase was even more pronounced when we stimulated T_reg cells from subjects with RRMS with IL-12 (Supplementary Fig. 1c), suggesting a higher intrinsic responsiveness to IL-12 in these individuals.

Figure 2.

Characterization of IL-12–driven, IFN-γ⁺Foxp3⁺ T_reg cells in vitro from healthy controls. (a) Intracellular staining for IFN-γ, IL-4, IL-17 and IL-10 of untreated (upper row) and IL-12-stimulated (bottom row) human T_reg cells from healthy controls at day 4. (b) mRNA expression of FOXP3 (left), GATA3 (middle) and TBX21 (right) in T_reg cells stimulated in the presence or absence of IL-12 for 5 d (data are a representative example of three experiments performed with similar results; *P < 0.05). (c) Proliferation of T_resp cells cocultured for 3 d with T_reg cells (T_reg cell:T_resp cell ratio of 1:2) previously treated with IL-2 (top) or IL-2 + IL-12 (bottom), as assessed by carboxyfluorescein succinimidyl ester (CFSE) dilution. Histograms depict unstimulated T_resp cells alone (top left), T_resp cells stimulated with antibody to CD3 (anti-CD3) and antigen-presenting cells without T_reg cells (bottom left) and T_resp cell and T_reg cell cocultures without blocking antibodies (second column), in the presence of an IL-10–specific blocking antibody (anti–IL-10; third column), IFN-γ–specific blocking antibody (anti–IFN-γ; fourth column) or both antibodies (fifth column). (d) Representative example of staining for T_H1-associated chemokines(CCR5 and CXCR3) and TGF-β on IFN-γ⁺Foxp3⁺ and IFN-γ⁻ Foxp3⁺ T_reg cells (data are representative of three experiments).

The capacity of T_reg cells from healthy subjects to secrete IFN-γ was not accompanied by loss of Foxp3 expression (Supplementary Figs. 1d and 3) and was dependent on the dose of IL-12 (Supplementary Fig. 4). IL-12 also induced a modest increase in IL-10, but not in IL-17 or IL-4, production (Fig. 2a and Supplementary Fig. 5). Other members of the IL-12 family of cytokines, including IL-23 and IL-27, did not induce either IFN-γ or IL-10 production (Supplementary Fig. 6). We confirmed at a single-cell level that IL-12 could induce IFN-γ and IL-10 production in T_reg cell clones (Supplementary Figs. 7 and 8).

As expected given the induction of IFN-γ secretion by IL-12, and similar to what we observed in ex vivo T_reg cells from subjects with RRMS IL-12–stimulated T_reg cells expressed significantly more TBX21 mRNA and protein and less GATA3 mRNA as compared to T_reg cells not treated with IL-12 (Fig. 2b and Supplementary Fig. 9). There was no change in the expression of FOXP3, RORC, IRF1 (encoding interferon regulatory factor 1) and MAF, a transcription factor related to IL-10 production (Fig. 2b and Supplementary Fig. 9). We obtained similar results with single-cell–derived T_reg cell clones stimulated for 10 d in the presence of IL-12 (Supplementary Fig. 10).

To examine whether IL-12–stimulated Foxp3⁺T-bet⁺ T_reg cells were functionally suppressive, we cocultured CD4⁺CD25^low/− responder T cells with T_reg cells that had been prestimulated for 4 d with IL-2 alone or with IL-2 and IL-12, along with antibody to CD3 and irradiated T cell–depleted peripheral blood mononuclear cells. After 3 d, T_reg cells cultured with IL-2 and IL-12 were significantly less effective at inhibiting responder T cell proliferation as compared to IL-2–treated T_reg cells (Fig. 2c and Supplementary Fig. 9c). Blocking IFN-γ increased the suppressive activity of T_reg cells cultured with IL-2 and IL-12 but not that of control cells treated with only IL-2. These data suggest that other mechanisms may also account for the defect in suppression. We obtained the same results in antigen-presenting cell–free coculture assays (Supplementary Fig. 11). We also observed diminished suppressive activity of IL-12–treated T_reg cell clones (Supplementary Fig. 12). Although our data point to a general defect in the suppressive activity of T_H1-like, Foxp3⁺ T cells, it is unclear whether these cells are able to specifically suppress T_H1 responses, as has been shown in a mouse model of inflammation¹³.

Next we examined the reversible nature of the T_H1-like T_reg cell phenotype. We collected T_reg cells cultured with IL-12 for 4 d and split them into two populations; one was washed to eliminate IL-12, whereas the other was cultured with IL-12 for 3 d longer. The frequency of IFN-γ⁺Foxp3⁺ T cells decreased when we cultured the cells without IL-12 for the last 72 h, and these T_reg cells reacquired suppressive activity (Supplementary Fig. 13). Viability staining of the populations excluded the possibility of selective apoptosis induced by the removal IL-12 (data not shown), suggesting that IL-12 did not induce permanent changes in the amount of IFN-γ expressed or the frequency of T_reg cells expressing IFN-γ.

IL-12–stimulated T_reg cells showed a T_H1 chemokine receptor profile^14,15 characterized by expression of both CXCR3 and CCR5 and decreased TGFB1 mRNA and protein expression (Fig. 2d and Supplementary Fig. 14). In contrast to T_reg cells from subjects with RRMS, T_reg cells upregulated CTLA4 mRNA and protein expression in response to IL-12 (Supplementary Fig. 15).

These data provide a general mechanism by which proinflammatory cytokines such as IL-12 can rapidly alter the phenotype and function of T_reg cells, decreasing their suppressive activity. Our results suggest one possible mechanism to account for the diminished suppressive activity of T_reg cells from individuals with multiple sclerosis^3–6, although prospective studies on the frequency of IFN-γ⁺Foxp3⁺ T cells in populations at risk for developing multiple sclerosis and in patients before and after treatment with IFN-β are needed to confirm whether these T_H1-like T_reg cells are associated with disease pathogenesis. Taken together, our results underscore the plasticity of T_reg cells in a proinflammatory environment.