Foxp3+ Treg cells in the inflamed CNS are insensitive to IL-6-driven IL-17 production


  • Richard A. O'Connor,

    1. Centre for Multiple Sclerosis Research and Centre for Immunology Infection and Evolution, Queen's Medical Research Institute, Medical Research Council/University of Edinburgh Centre for Inflammation Research, Edinburgh, UK
    Search for more papers by this author
  • Stefan Floess,

    1. Experimental Immunology, Helmholtz Centre for Infection Research, Braunschweig, Germany
    Search for more papers by this author
  • Jochen Huehn,

    1. Experimental Immunology, Helmholtz Centre for Infection Research, Braunschweig, Germany
    Search for more papers by this author
  • Simon A. Jones,

    1. Department of Infection, Immunity & Biochemistry, The School of Medicine, Cardiff University, Cardiff, Wales, UK
    Search for more papers by this author
  • and Stephen M. Anderton

    Corresponding author
    1. Centre for Multiple Sclerosis Research and Centre for Immunology Infection and Evolution, Queen's Medical Research Institute, Medical Research Council/University of Edinburgh Centre for Inflammation Research, Edinburgh, UK
    • Prof. Stephen M. Anderton, Centre for Inflammation Research, Queen's Medical Research Institute, University of Edinburgh, 47 Little France Crescent, Edinburgh, EH16 4TJ, UK

      Fax: +44-131-242-6560


    Search for more papers by this author


Foxp3+ T regulatory (Treg) cells can be induced to produce interleukin (IL)-17 by in vitro exposure to proinflammatory cytokines, drawing into question their functional stability at sites of inflammation. Unlike their splenic counterparts, Treg cells from the inflamed central nervous system (CNS-Treg cells) during EAE resisted conversion to IL-17 production when exposed to IL-6. We show that the highly activated phenotype of CNS-Treg cells includes elevated expression of the Th1-associated molecules CXCR3 and T-bet, but reduced expression of the IL-6 receptor α chain (CD126) and the signaling chain gp130. We found a lack of IL-6 receptor on all CNS CD4+ T cells, which was reflected by an absence of both classical and trans-IL-6 signaling in CNS CD4+ cells, compared with their splenic counterparts. We propose that extinguished responsiveness to IL-6 (via down-regulation of CD126 and gp130) stabilizes the regulatory phenotype of activated Treg cells at sites of autoimmune inflammation.


Foxp3+ Treg cells are primary mediators of peripheral tolerance and have shown therapeutic potential in models of organ-specific autoimmune disease [[1]]. However, Treg cells have also been reported to produce interleukin (IL)-17 when stimulated in vitro in the presence of inflammatory cytokines [[2, 3]], suggesting that Treg cells can adapt to an inflammatory environment by acquiring certain effector characteristics. Here, we tested whether Treg cells isolated from a site of autoimmune inflammation could be driven toward an effector phenotype. We used the experimental autoimmune encephalomyelitis (EAE) model wherein Foxp3+ Treg cells accumulate in the inflamed central nervous system (CNS). Unlike their splenic counterparts, CNS-Treg cells resisted conversion into an IL-17-secreting population. This resistance was attributable to a reduction in IL-6 responsiveness due to the fact that CNS-Treg cells lacked expression of both chains of the IL-6 receptor, CD126, and gp130. We therefore reveal a key mechanism allowing Treg cells that are active in sites of inflammation to maintain a commitment to an antiinflammatory role.

Results and discussion

CNS-Treg cells resist conversion to an IL-17-producing phenotype

We fluorescence-activated cell sorter (FACS)-sorted Treg (GFP+) and non-Treg (GFP) CD4+ cells from the spleen and CNS of Foxp3-GFP mice with EAE and assessed their cytokine production profile. CNS Foxp3 T cells showed production of IL-2 and a broad range of effector cytokines (IL-4, IL-5, IL-17, IFN-γ, TNF-α, and GM-CSF) in response to anti-CD3+anti-CD28 stimulation. In contrast, Foxp3+ cells from the CNS showed no production of these effector cytokines, with only low-level production of IL-10 being evident (Fig. 1A). We next tested FACS-sorted GFP+ (Foxp3+) CNS-Treg cells under in vitro exposure to a well-characterized IL-17-promoting cocktail. Consistent with earlier reports [[3]], around 7% of splenic Treg cells from naïve mice produced IL-17 in response to T-cell receptor for antigen stimulation in the presence of IL-1β, TGF-β, IL-6, and IL-23 (Fig. 1B). Splenic Treg cells from mice with EAE produced IL-17 at a similar frequency, indicating that there was no systemic perturbation in the capacity of Treg cells to produce IL-17 during EAE. However, the frequency of IL-17+ cells was markedly lower in the Treg-cell population sampled from the inflamed CNS of those same mice with EAE (Fig. 1B and C) and was reflected in the level of IL-17 detected in these cultures (Fig. 1D). As Th1-associated effector cytokines act as negative regulators of Th17 differentiation, we tested whether CNS-Treg cells produced IFN-γ, but found no evidence for this under any conditions tested, including exposure to IL-12 (Supporting Information Fig. 1). Bisulphite sequence analysis of CpG motifs within the Treg-specific demethylation region (TSDR) revealed complete demethylation in both splenic and CNS-Treg cells (Fig. 1E), a pattern associated with natural Treg cells rather than the incomplete demethylation seen among in vitro generated iTreg cells [[4]]. Therefore, epigenetic differences at the TSDR did not account for the inability of CNS-Treg cells to produce IL-17. Previous studies have shown that the increased proportion of Foxp3+ T cells in the CNS during EAE is not due to the peripheral conversion of Foxp3 T cells to Foxp3+ adaptive Treg cells [[5]]. Our analysis of the TSDR supports this view.

Figure 1.

CNS-Treg cells resist conversion to an IL-17 producing phenotype. (A) Cytokine production by GFP and GFP+ CD4+ T cells from the CNS of Foxp3-GFP mice with EAE. Supernatants were tested after 72 h stimulation with plate-bound anti-CD3 and anti-CD28. (B) IL-17-production by FACS-sorted GFP+ (Foxp3+) splenic and CNS-Treg cells from mice with EAE and splenic Treg cells from naïve mice, stimulated for 3 days in the presence of the indicated cytokines, prior to restimulation with PMA and ionomycin. (C) Percentage of Treg cells producing IL-17. (D) IL-17 production by Treg cells as measured by ELISA. (B, C, D) Data are shown as mean + SEM of triplicate wells from one experiment representative of three with 12–15 mice per group in each experiment. (E) Bisulphite sequencing of the TSDR in splenic and CNS-derived populations of sorted GFP and GFP+ cells from Foxp3-GFP mice during recovery from EAE (n = 30).

CNS T cells show reduced responsiveness to IL-6 and express low levels of CD126 and gp130

IL-6 can drive IL-17 production by naïve T cells and by Treg cells [[2, 6]]. The IL-6 receptor is composed of an IL-6-specific α chain (CD126) coupled with the signaling chain gp130, which is shared with other cytokine receptors (reviewed in [[7]]). Cells lacking surface expression of the IL-6R can also respond to IL-6 bound to the soluble form of the IL-6Rα, which then binds gp130 at the cell surface to provide IL-6 trans-signaling [[8]]. Peripheral Foxp3 and Foxp3+ T cells from naïve mice responded rapidly to either IL-6 or hyper DS s-IL-6R (HDS), an IL-6-sIL-6R fusion protein that triggers trans-signaling [[9]], as measured by the appearance of pSTAT1 and pSTAT3 (Supporting Information Fig. 2). However, unlike their splenic counterparts, CNS CD4+ cells from mice with EAE showed no expression of pSTAT1 or pSTAT3 after incubation with either IL-6 or HDS (Fig. 2A). Notably, this insensitivity was evident on all CNS CD4+ cells and was not restricted to the Treg-cell population.

Figure 2.

CNS-infiltrating T cells show reduced responsiveness to IL-6 and express low levels of CD126 and gp130. (A) Levels of phosphorylated STAT1 and STAT3 in CD4+ T cells from the spleen (upper panels) or CNS (lower panels) of mice with EAE, in response to incubation with IL-6 or HDS. (B) Expression of CD126 on splenic and CNS CD4+ T cells is shown as mean + SEM of n = 5. (C) gp130 expression in splenic and CNS CD4+ T cells taken from Foxp3-GFP mice with EAE. Filled histograms indicate staining with an isotype-matched control antibody. Mean fluorescence intensity data in the right panel are compiled showing individual mice (n = 7). Data are from one of three experiments giving consistent results.

The relative resistance of induced Treg cells to the induction of IL-17 production has been correlated with their loss of IL-6 receptor expression [[10, 11]]. Reduced CD126 expression on CNS CD4+ cells would account for their insensitivity to IL-6, but they would be predicted to maintain responsiveness to IL-6 trans-signaling if they still expressed gp130. We found that both GFP+ and GFP CD4+ cells from the CNS showed markedly reduced levels of both CD126 and gp130 in comparison with their splenic counterparts from the same mice (Fig. 2B and C).

A subset of Treg cells in naive mice expresses T-bet and CXCR3 which enhance their migration to sites of Th1-driven inflammation [[12]]. During EAE, IFN-γ drives local expression of CXCL10, a ligand for CXCR3, in the inflamed CNS [[13]]. CNS T cells showed elevated expression of T-bet and CXCR3 which was particularly high in CNS-Treg cells (Fig. 3A). CXCR3 expression correlated with the absence of CD126 on CD4+ cells from naïve spleen (Fig. 3B) suggesting that the CXCR3+ Treg cells which arrive at the CNS early after the onset of inflammation will be drawn from a pool mostly lacking CD126 expression.

Figure 3.

CD126-splenic Treg cells resist conversion to IL-17 production. (A) The expression of CXCR3 and T-bet by splenic (left) and CNS-infiltrating (right) CD4+ T cells isolated at the peak of disease (13 days post immunization) is shown. (B) Representative CD126 and CXCR3 expression profiles of Foxp3+ and Foxp3 CD4+ cells isolated from the spleen of naive mice are shown. (C) IL-17 production by CD126+ and CD126 GFP+ and GFP CD4+ T cells sorted from naïve Foxp3-GFP mice and stimulated on plate bound anti-CD3 and anti-CD28 for 3 days without addition of exogenous cytokines (medium), or in the presence of TGF-β, IL-1β, IL-23, with or without addition of IL-6, is shown. Data are representative of two experiments giving consistent results.

CD126-splenic Treg cells resist conversion to IL-17 production

The model that develops from these data is that, in vivo, Treg cells might be susceptible to IL-6-driven diversion to an IL-17-producing phenotype when expressing CD126 and gp130 (i.e. in the lymphoid organs, as can be seen by the ability of splenic Treg cells from mice with EAE to produce IL-17 upon in vitro exposure to an IL-6-containing cocktail (Fig. 1B). However, upon arrival in the organ under autoimmune attack, Treg cells have lost this capacity because they have down-regulated CD126 and gp130. Of course, this loss of receptors was not restricted to Treg cells; they were also low/absent on CNS GFP cells (Fig. 2B and C) and pSTAT1 and pSTAT3 were absent in all CNS CD4+ cells exposed to either IL-6 or HDS. However, CNS GFP cells (but not GFP+ cells) are clearly able to produce large quantities of IL-17 (Fig. 1A). This is most likely maintained because effector cells, initially triggered in the presence of IL-6, are induced to express the IL-23R [[14]]. IL-23 is readily available in the inflamed CNS during EAE [[15]], but the IL-23R is not expressed by Treg cells [[16]]. Therefore, we propose that although both CNS T effectors and Treg cells are insensitive to IL-6 signaling, their differential sensitivity to IL-23 allows T effectors to maintain IL-17 production. Lack of CD126 should therefore serve as a marker of preactivated Treg and T effectors. We sorted splenic GFP+ and GFP cells, that either did or did not express CD126, from naïve Foxp3-GFP mice and found that CD126+ cells produced IL-17 only if IL-6 was included in the culture while GFPCD126 cells would produce IL-17 in IL-23-containing medium without IL-6 (Fig. 3C). Furthermore, GFP+CD126 cells could not be provoked to produce IL-17, consistent with the reported absence of IL-23R from Treg cells [[16]].

CNS-Treg cells express T-bet, CXCR3 and have lost CD126 (Fig. 3). Expression of CXCR3 is T-bet dependent [[12]]. However, CXCR3 expression was not a surrogate marker identifying IL-6-insensitive Treg cells. Sorted CXCR3+ splenic Treg cells from naïve mice maintained the ability to produce IL-17 (Supporting Information Fig. 3), correlating with ∼20% of Foxp3+CXCR3+ cells expressing CD126 (as shown in Fig. 3B). We also found that sorted naïve Treg cells down-regulated CD126 efficiently after anti-CD3/anti-CD28 stimulation and that this down-regulation was not enhanced by exposure to IL-12, which drives T-bet expression in Treg cells (data not shown). Cumulatively, these data therefore suggest that the inability to respond to IL-6 is not a direct consequence of T-bet expression by Treg cells.

Exposure to retinoic acid (RA) promotes resistance to IL-17 production in nTreg via down-regulation of CD126 expression [[17]]. RA is produced at sites of inflammation [[18]] and whether such an effect in the inflamed CNS might maintain the IL-6-insensitive phenotype of CNS T cells is worthy of further investigation.

Concluding remarks

Recent fate-mapping studies showed that the majority of CD4+ effector T cells infiltrating the CNS during EAE have, at some point, produced IL-17 [[19, 20]]. Unlike their Foxp3 counterparts however, CNS-derived Foxp3+ cells showed no history of IL-17 expression [[20]]. We can therefore conclude that the inflammatory environment within the CNS fails to induce IL-17 production by the infiltrating Foxp3+ T cells and, from our data here, that these cells resist conversion, even when experimentally challenged under potent IL-17-inducing conditions that work on Treg cells taken from noninflamed sites.

Besides inducing IL-17 production in Treg cells, several inflammatory cytokines, including IL-6, can also render effector T cells resistant to suppression as measured using in vitro assays [[5, 21]]. On this point, our data on the insensitivity of CNS GFP cells to IL-6 are noteworthy, and would exclude such a function of IL-6 within the CNS, at least one that acted directly on T cells.

We demonstrate that the response of CNS-Treg cells to inflammatory cytokines cannot be predicted accurately from the behavior of peripheral Treg cells taken from the same individual. This has implications for human studies that sample Treg cells from the circulation, such as the recent description of elevated IFN-γ production by peripheral blood Foxp3+ cells from multiple sclerosis (MS) patients [[22]]. The prediction from our study would be that CNS-Treg cells in MS might maintain suppressive, rather than effector function. Furthermore, concerns that Treg cells that have been manipulated therapeutically might develop unwanted effector function (based on in vitro observation using “naïve” Treg cells) might be overstated.

Perhaps the most interesting feature of our current comparison of CNS and peripheral T cells is the apparent loss of gp130 from all CD4+ cells in the CNS, given that gp130 is the signaling unit for other cytokines, including IL-11, IL-27, and leukemia inhibitory factor (reviewed in [[7]]). The down-regulation of gp130 should render CNS T cells insensitive to the effects of these cytokines also. Spatial and temporal variation in the expression of cytokine receptors therefore offers a fundamental means of controlling effector and Treg-cell function at different stages of an inflammatory immune response. This possibility certainly warrants further study.

Materials and methods

Mice, antigens, and EAE

Foxp3-GFP mice [[23]] and Foxp3.LuciDTR-4 mice [[24]] were bred under specific pathogen-free conditions at the University of Edinburgh. All experiments were approved by the University of Edinburgh ethical review committee and were performed in accordance with UK legislation. The 35–55 peptide of myelin oligodendrocyte glycoprotein (pMOG) was obtained from Cambridge Research Biochemicals. EAE was induced using 100 μg of pMOG and mononuclear cells were prepared from brain and spinal cord as described previously [[25]].

In vitro culture conditions and cytokine measurement

GFP+ or GFP-CD4+ T cells were sorted using a FACSAria II sorter (BD Biosciences, Oxford, UK). Purities were routinely greater than 99%. Cells were stimulated on anti-CD3 + anti-CD28 (e-Bioscience, CA, USA) coated plates, with or without IL-6 (30 ng/mL), IL-23 (30 ng/mL), IL-1β (10 ng/mL), TGF-β (2.5 ng/mL), or IL-12 (25 ng/mL) (all R&D systems), individually or in combination, as described in the text. Cytokine production was quantified using ELISA or Bender-Medsystems FLowcytomix Th1/Th2 10plex assays (e-Bioscience,) according to the manufacturer's instructions.

Antibodies and FACS analysis

All antibodies were from e-Bioscience, except pSTAT1, pSTAT5, and pSTAT3 (BD Pharmingen, Oxford, UK). For intracellular cytokine staining, 50 ng/mL PMA, 50 ng/mL ionomycin, and 1 μL/mL brefeldin A (e-Bioscience) were added for the last 4 h of culture. Foxp3 staining was performed using proprietary buffers according to the manufacturer's instructions (e-Bioscience). Due to loss of GFP activity as a result of fixation, cells from Foxp3.LuciDTR-4 mice were stained with anti-Foxp3. For pSTAT analysis, cells were incubated in RPMI 10% FCS with or without IL-6, or the sIL-6R-IL-6 fusion protein HDS [[26]], both at 20 ng/mL for 15 min at 37°C and fixed in 2% PFA for 20 min at 37°C prior to surface staining. Cells were then resuspended in ice-cold 90% methanol and stored overnight at −20°C. Cells were then washed extensively and incubated with Fc block before intracellular staining. All FACS data were analyzed using FlowJo software (Tree Star, CA, USA). Statistical analysis used Student's t-test for comparison of groups.

TSDR methylation analysis

Genomic DNA was isolated from freshly sorted cells using a DNeasy blood and tissue kit (Qiagen, Crawley, UK) according to the manufacturer's instructions. Bisulfite conversion, PCR, and sequencing was performed as previously described [[4]].


We thank Prof. A. Rudensky for providing the Foxp3-GFP mice and Prof. G. Hammerling for providing the Foxp3.LuciDTR-4 mice. This work was supported by grants from the UK Medical Research Council and the German Research Foundation (SFB621 and KFO250).

Conflict of interest

The authors declare no financial or commercial conflicts of interest.


hyper DS s-IL-6R


treg-specific demethylation region