A Th17-like developmental process leads to CD8+ Tc17 cells with reduced cytotoxic activity

Authors


Abstract

Activation of naive CD8+ T cells with antigen in the absence of skewing cytokines triggers their differentiation into effector CTL, which induces death of target cells. We show that CD8+ T cells activated in the presence of the cytokines IL-6 or IL-21 plus TGF-β similar to CD4+ T cells, develop into IL-17-producing (Tc17) cells. These cells display greatly suppressed cytotoxic function along with low levels of the CTL markers: T-box transcription factor Eomesodermin, granzyme B and IFN-γ. Instead, these cells express hallmark molecules of Th17 program including retinoic acid receptor-related orphan receptor (ROR)γt, RORα, IL-21 and IL-23R. The expression of the type 17 master regulator RORγt is causally linked to Tc17 generation, because its overexpression stimulates production of IL-17 in the presence of IL-6 or IL-21. Both, upregulation of the type 17 program as well as suppression of CTL differentiation are STAT3 dependent. Furthermore, Tc17 cells producing IL-17 but not granzyme B are also detectable in EAE, a mouse model for multiple sclerosis. Our data point to the existence of mutually exclusive CTL and Tc17 developmental pathways in vitro and in vivo.

Introduction

CTL are important effector cells in the immune response to intracellular pathogens and tumors. They differentiate from naive CD8+ T cells following activation by antigen in the absence of skewing cytokines, and during this process they acquire the ability to destroy their targets by releasing cytotoxic molecules such as perforin and granzymes, from granules into the immunological synapse. In addition, CTL secrete cytokines, mostly IFN-γ and TNF-α, which function to induce or augment inflammation 1–3.

Two T-box transcription factors, Eomesodermin (Eomes) and T-bet, are important for the development of effector and memory CTL 4–6. Studies using deletion, overexpression or dominant negative analogs of these factors have suggested that both of them are involved in the regulation of expression of IFN-γ, granzyme B and perforin 4, 5, 7. Consistent with these data, CD8+ T cells with combined deletion of the Eomes and Tbx21 (encoding T-bet) genes differentiate into cells with highly impaired cytotoxic activity and IFN-γ production 8. Instead, these cells produce Th17 type cytokines and express the IL-23 receptor (IL-23R) as well as the transcription factor retinoic acid receptor-related orphan receptor (ROR)γt, both of which are characteristic for the type 17 differentiation program 8. Thus, the phenotype of CD8+ T cells deficient for both Eomes and T-bet is reminiscent of the newly described Th17-cell subset.

Th17 cells produce IL-17A, IL-17F, IL-21 and IL-22, which are highly pro-inflammatory and induce severe autoimmunity, e.g. during EAE, the mouse model for multiple sclerosis 9. The differentiation of these cells requires TGF-β in combination with IL-6 10, 11. Two additional cytokines, IL-21 and IL-23, are also critically involved in the differentiation of Th17 cells 11, 12. In combination with TGF-β, IL-21 induces and amplifies Th17 development independently of IL-6 11, 13, 14. IL-21 further induces its own expression in an autocrine manner and additionally upregulates the IL-23R 11, 13, 14. Thus, Th17 differentiation is induced by antigenic stimulation and the sequential action of the cytokines IL-6, IL-21 and IL-23, along with TGF-β.

The development of Th17 cells is orchestrated by two STAT3-regulated lineage-specific transcription factors, RORγt and RORα. These transcription factors are sufficient to cooperatively augment the expression of IL-17A, IL-17F and IL-23R 15. Accordingly, loss of function in either RORα or RORγt leads to a reduction, whereas the combined mutation of both molecules profoundly impairs Th17 differentiation in vitro and in vivo 15, 16. In addition, our group has shown that the IFN regulatory factor 4 is mandatory for IL-6 as well as IL-21-induced Th17 differentiation 17, 18.

In the current study, we show that the differentiation of CD8+ T cells in the presence of the cytokines IL-6 and IL-21 along with TGF-β induces very low amounts of typical CTL expressing granzyme B, Eomes and IFN-γ. Instead, these cytokines induce the development of IL-17 positive cells (Tc17) that display a type 17 profile with high IL-21, IL-23R, RORα and RORγt. Consistent with the impaired expression of molecules characteristic for CTL, these cells have greatly impaired killing activity as compared with CTL. We further show that development of these cells depends on STAT3 and can be augmented by overexpression of RORγt. Finally, Tc17 cells producing IL-17 but not granzyme B are detectable in vivo during EAE. Thus, CTL and Tc17 effectors arise predominantly in a mutually exclusive fashion in vitro and in vivo.

Results

Mutual exclusive expression of IL-17 and granzyme B or IFN-γ in effector CD8+ T cells

To elucidate, whether WT CD8+ T cells are able to acquire the type 17 phenotype, we chose culture conditions normally used for induction of Th17 cells 10, 11, 13, 14. We stimulated purified CD8+ T cells with plate-bound anti-CD3 and soluble anti-CD28 antibodies in the presence or absence of IL-6, IL-21 and TGF-β alone, or in combination for 3 days. After restimulation, the cells were stained intracellularly for IL-17, the hallmark cytokine of the type 17 fate 9, and for granzyme B and IFN-γ, the attributes of effector CTL 1, 3. Then, the cells were assessed by flow cytometry. As expected, in the absence of skewing cytokines, CD8+ T cells developed the typical effector CTL phenotype with high expression of the central cytotoxic molecule granzyme B and the cytokine IFN-γ (Fig. 1A). Cells with the highest level of IFN-γ uniformly co-produced granzyme B (Fig. 1B). The cytokines IL-6 or IL-21 on their own did not influence granzyme B, but they markedly suppressed IFN-γ production (Fig. 1A and B). In contrast, addition of TGF-β to the differentiating cells inhibited both granzyme B and IFN-γ production and caused an almost complete loss of double positivity for these molecules (Fig. 1A and B). Notably, similar to CD4+ T cells, also in CD8+ T cells IL-17 production was induced by the cytokines IL-6 or IL-21 in the presence of TGF-β (Fig. 1A and B). IL-17 production corresponded to decreased expression of granzyme B and IFN-γ (Fig. 1A).

Figure 1.

Analysis of cytokine and granzyme B expression as well as cytolytic function of CTL and Tc17 cells. (A and B) Flow cytometry of purified CD8+ T cells differentiated for 72 h by stimulation via CD3/28 in the absence (CTL) or presence of the indicated cytokines, restimulated and stained intracellularly for IL-17, granzyme B, or IFN-γ. Numbers indicate the frequencies of IL-17+, IFN-γ+, granzyme B+ cells and of cells double positive for granzyme B and IFN-γ. (C and D) Antigen-specific cytolytic activity of OT1 CTL or Tc17 cells. Purified OT1 cells were differentiated for 96 h under conditions as in (A) and then added at the indicated ratios to CFSE labeled target EL4 cells either pulsed with SIINFEKL peptide or not. Percent of specific lysis was determined after 6 h. Data (±SD) represent triplicate determinations. Representative data from three independent experiments for A–D are shown.

This mostly exclusive expression of IL-17 and either granzyme B or IFN-γ suggested that culture of CD8+ T cells with the cytokines IL-6 or IL-21 along with TGF-β skewed differentiation toward the type 17 fate. Hence, the arising cells might have distinct functional characteristics in addition to their differing phenotype. Therefore, we next compared their cytotoxic activity. For this, we stimulated OT1 cells with plate-bound anti-CD3 and soluble anti-CD28 antibodies in the presence or absence of IL-6, IL-21 and TGF-β alone or in combination for 96 h. Similar to WT CD8+ T cells, IL-17 production of OT1 cells was associated with decreased expression of granzyme B and IFN-γ (Supporting Information Fig. 1). Antigen-specific effector CTL activity was quantified by the detection of CFSE-stained EL4 thymoma cells pulsed with SIINFEKL peptide. Although IL-6 or IL-21-treated CD8+ T cells expressed IFN-γ at diminished levels (Fig. 1A, Supporting Information Fig. 1), the cytolytic capacity of these cells was not significantly influenced in accordance with their unaltered levels of granzyme B (Fig. 1A–C, Supporting Information Fig. 1). We were able to confirm the previously described enhancing effects of IL-6 or IL-21 on the killing activity 19, 20, but only when we used suboptimal CTL differentiation conditions without IL-2 (Supporting Information Fig. 2).

As expected from the suppressed granzyme B expression (Fig. 1A, Supporting Information Fig. 1), the cytolytic capacity of OT1 cells cultured with TGF-β in combination with IL-6 or IL-21 was greatly diminished (Fig. 1D). This impaired killing activity cannot be explained by defective proliferation as described for CD8+ T cells lacking the transcription factor Runx3 21, because cell division of CD3/28 stimulated CFSE-labeled cells was similar after culture in the presence or absence of IL-6, IL-21 or TGF-β alone, or in combination (Supporting Information Fig. 3).

Thus, the reciprocal expression of IL-17 and granzyme B in CD8+ T cells is reflected in the greatly diminished cytotoxic function of Tc17 cells as compared with CTL.

Molecular characterization of Tc17 cells

To establish the transcriptional program of CD8+ T cells cultured under the above-described conditions (Fig. 1), we performed quantitative RT-PCR analysis for the hallmark molecules of type 17 and CTL differentiation. As already mentioned in the introduction, besides IL-17 production, Th17 cells are characterized by the expression of IL-21 that in an autocrine manner amplifies their development and additionally upregulates the expression of the IL-23R. Further, the IL-23R is necessary for receiving the stabilizing signals mediated by IL-23 22. Both IL-17 production and IL-23R expression are dependent on two master transcription factors, RORα and RORγt (encoded by Rora and Rorc, respectively), which cooperatively orchestrate Th17 development 15, 16. In contrast, the CTL program is characterized by the expression of Eomes, the critical transcription factor in directing the lytic effector development of CD8+ T cells and the cytokine IFN-γ4. Therefore, we evaluated in differentiating CD8+ T cells expression of Il21, Il23r, Rora and Rorc mRNA on the one hand, and of Eomes and Ifng on the other. Similar to their effects on CD4+ T cells 11, 23, IL-6 or IL-21 alone induced robust expression of Il21 and Il23r in differentiating CD8+ T cells (Fig. 1A) and at the same time downregulated Eomes and Ifng (Fig. 1B). This result is consistent with a published inhibitory effect of IL-21 on expression of CTL marker molecules 24. Addition of TGF-β suppressed the expression of Il23r, an effect already described for CD4+ T cells 23, but did not influence Il21 expression (Fig. 2A). As described for CD4+ T cells 11, 13–16, both key transcription factors Rora and Rorc were strongly upregulated in differentiating CD8+ T cells by the combination of IL-6 or IL-21 plus TGF-β (Fig. 2A). Concomitantly, Eomes and Ifng were significantly suppressed (Fig. 2B).

Figure 2.

Transcriptional profile and function of RORγt in CTL and Tc17 cells from WT mice. (A and B) Real-time RT-PCR for Eomes, Ifng, Il21, Il23r, Rora and Rorc mRNA in purified CD8+ T cells stimulated for 40 h via CD3/28 under the indicated cytokine conditions. The respective expression levels in CD8+ cells under CTL conditions (for A), or in the presence of IL-6 plus TGF-β (for B) were set to one. Data represent duplicate determinations. (C) Flow cytometry of purified CD8+ T cells transduced with retroviruses expressing either RORγt-GFP or GFP alone, cultured in the presence of the indicated cytokines, and stained for intracellular IL-17. The numbers indicate the percentages of IL-17 positive cells that are related to the non-transduced GFP cells (top left) or to the transduced GFP+ cells (top right). Representative data from three independent experiments for A–C are shown.

Similarly to Th17 cells, we observed a significant induction of mRNA for RORγt in CD8+ T cells cultured under IL-17 inducing conditions (Fig. 2A). To analyze the causal relation between RORγt expression and IL-17 production in CD8+ T cells, we examined the effect of overexpression of RORγt in differentiating CD8+ T cells. Therefore, purified CD8+ T cells were transduced with retroviruses expressing RORγt-GFP or GFP only. After infection, the cells were cultured in the presence or absence of IL-6 or IL-21 alone, or in the combination with TGF-β. After restimulation, IL-17-producing cells were assessed by intracellular staining. RORγt was sufficient to induce high levels of IL-17 production in CD8+ T cells differentiating in the presence of IL-6 or IL-21. In contrast, in the absence of these cytokines the overexpression of RORγt induced very low frequencies of IL-17 positive cells (Fig. 2C) This result demonstrates that in developing CTL, RORγt is able in limited extent to induce type 17 differentiation. This is probably caused by an antagonizing function of Eomes on RORγt; this conclusion is derived from findings in STAT6/T-bet doubly deficient CD4+ T cells in which forced expression of Eomes suppressed IL-17 production 25. Indeed, in the presence of IL-6 or IL-21 alone when Eomes mRNA expression is suppressed (Fig. 2B), overexpression of RORγt was sufficient to strongly induce IL-17-producing cells (Fig. 2C). Therefore, these results suggest that RORγt is a critical factor in directing IL-17 production in CD8+ T cells.

In summary, these data demonstrate that CD8+ T cells cultured in the presence of IL-6 or IL-21 and TGF-β undergo the type 17 developmental program characterized by the production of IL-17, expression of Il21, Il23r and the lineage-specific transcription factors Rora and Rorc. Concomitantly, these cells express Eomes, granzyme B and IFN-γ at greatly diminished levels. Moreover, RORγt expression can be causally linked to IL-17 production in CD8+ T cells. Our results suggest a dichotomy in the generation, phenotype, function and dependency on lineage-specific transcription factors of CTL and Tc17 cells. Hence, Tc17 cells can be defined as a distinct CD8+ T-cell subset with unique phenotypical and functional characteristics.

Tc17 development depends on STAT3

A common feature of IL-6- and IL-21-mediated intracellular signal transduction is the activation of STAT3 22, which is necessary for the expression of the type 17 master regulators RORα and RORγt 15, 26, 27 Consequently, the Th17 developmental program is greatly compromised in STAT3-deficient cells 11, 26, 27. To evaluate whether similar to Th17, STAT3 is involved in regulating Tc17 development, we transiently downregulated the expression of STAT3 in CD8+ T cells using siRNA. To achieve this, we nucleofected purified CD8+ WT cells with two different siRNA constructs, with either siRNA-1 (si1) or with siRNA-2 (si2) targeting STAT3. As a control, we used scrambled siRNA (Scr) to exclude any unspecific influence of the nucleofection protocol on the development of Tc17 cells. Forty hour after nucleofection, immunoblotting was performed and lower STAT3 protein amounts was demonstrated in cells nucleofected with si1 or si2 compared with control cells nucleofected with Scr (Fig. 3A). This transient knock-down was specific for STAT3, because the level of STAT1 was not affected by both siRNA (Fig. 3A). The si2 construct was significantly more efficient in inhibiting STAT3 expression and function than si1 (Fig. 3A–E). Functionally, the knock-down caused a significant reduction in the frequency of IL-17-producing cells induced by IL-6 or IL-21 in combination with TGF-β, as compared with control nucleofection with Scr (Fig. 3B and C). These results indicate a key role of STAT3 in the regulation of IL-17 production in CD8+ T cells. Conversely, the proportion of cells expressing the CTL marker molecules granzyme B and IFN-γ was increased when STAT3 was downregulated, suggesting a role for STAT3 in mediating repression of the CTL program by IL-6 and IL-21 signaling. To elucidate further the role of STAT3 in the regulation of Tc17 differentiation, we compared the mRNA profile of the silenced cells for type 17- and CTL-hallmark molecules. In keeping with the staining data and similar to the role of STAT3 in Th17 development 11, 26, the expression of the type 17 markers Il21, Il23r, Rora and Rorc, was inhibited in cells with transient STAT3 knock-down (Fig. 3D and E). Conversely, in these cells the expression of the CTL-hallmark molecules Eomes and Ifng was elevated (Fig. 3D and E).

Figure 3.

The function of STAT3 during Tc17 differentiation. CD8+ WT cells were purified and immediately nucleofected either with control Scr, or with si1 or si2 siRNA molecules targeting STAT3 and further cultured. (A) Immunoblot of STAT1, pSTAT3, STAT3 and β-actin 40 h after nucleofection. (B and C) Flow cytometry of nucleofected cells after priming for 72 h with IL-6 or IL-21 plus TGF-β and restimulation. The percentages of IL-17+, IFN-γ+ and granzyme B+ cells are indicated. (D and E) Real-time RT-PCR for Eomes, Ifng, Il21, Il23r, Rora and Rorc mRNA in nucleofected cells stimulated for 40 h via CD3/28 in the presence of the indicated cytokines. For Eomes and Ifng, the expression level in cells nucleofected with si2 was set to 1. For the other genes, the expression level in cells nucleofected with Scr was set to 1. Data represent duplicate determinations. Representative data from three independent experiments for A–E are shown.

These data establish the key role of STAT3 in directing the Tc17 program and the molecular switch between Tc17 and CTL development after treatment of CD8+ T cells with IL-6 or IL-21 in concert with TGF-β.

Tc17 cells are detectable in vivo

Th17 cells are critical for inflammation and disease in the CD4+ T–cell-mediated EAE 9. To analyze whether Tc17 cells are detectable in vivo, we used an EAE model in which CD8+ T cells have been shown to be involved 28. For this, we immunized mice with myelin oligodendrocyte glycoprotein (MOG) peptide MOG 37–50 emulsified in CFA. This protocol has been demonstrated to induce EAE characterized by the presence of peptide-specific CD8+ T cells in the CNS prior to and during the disease 28. As previously described 28, the MOG 37–50 immunized mice developed severe EAE (Fig. 4A, Table 1). To determine IL-17, IFN-γ and granzyme B production by CD8+ T cells from MOG-treated mice we isolated cells from draining LN and CNS at the induction phase or peak of the disease, respectively. The cells were stimulated with PMA and ionomycin and analyzed by flow cytometry. IL-17-producing CD8+ T cells were readily detectable in the analyzed organs at the induction phase and peak of disease (Fig. 4B and C). Although a minority of the IL-17-producing CD8+ T cells were also positive for IFN-γ (Fig. 4B and C), none of them costained for the cytotoxic effector molecule granzyme B (Fig. 4B and C). These results reinforce the concept that CD8+ T cells producing IL-17 are distinct from CTL. Consistent with a very recently published report 29, we were also able to detect Tc17 cells in the lung during influenza A infection in mice (data not shown).

Figure 4.

Detection of Tc17 cells during EAE. (A) Clinical signs of EAE (mean EAE score±SEM) induced by immunization of C57BL/6 mice (n=6) with MOG37–55 peptide. One representative course of five independent experiments is shown. (B and C) Intracellular cytokine staining of IL-17 and IFN-γ in CD8+ T cells. Lymphocytes were isolated from draining LN of mice within the induction phase of EAE (day 8 post immunization) or from the CNS of mice with acute EAE (27 dpi), respectively, and were restimulated in vitro with PMA and ionomycin. Viable CD8+ T cells are shown; numbers in quadrants indicate frequencies (%). Representative data from two independent experiments for B and C are shown.

Table 1. Clinical parameters of EAE induced by MOG37–50
EAE incidenceDay of onset mean (range)Maximal EAE score mean (range)MortalityDay of death/Sacrifice (range)
  1. a

    C57BL/6 mice [n=6] were immunized with MOG37–50 peptide in CFA and clinical signs of EAE were daily assessed. Indicated are the mean values as well the range of individual data of one experiment. Representative data from five independent experiments are shown.

100% (6/6)12.7 (12–15)4.7 (3–5)83.3% (5/6)23.4 (20–27)

Altogether, these data show the presence of IL-17-producing CD8+ T cells in vivo during a pathological process and are consistent with a reciprocal relationship between CTL and Tc17 phenotype, in which the IL-17-producing CD8+ T cells are negative for the cytotoxic molecule granzyme B.

Discussion

IL-17-producing CD8+cells have been described in several models in vivo using gene deficient or WT mice. In mice with a combined mutation in T-bet and Eomes, CD8+ T cells manifest poor cytotoxic activity, express a type 17 differentiation program and fail to control acute infection with lymphocytic choriomeningitis virus 8. In T-bet-deficient mice, IL-17-producing CD8+ cells are responsible for an allograft rejection, which is resistant to a blockade of costimulatory signals 30. A different study in T-bet-deficient mice described experimental autoimmune myocarditis caused by IL-17-producing CD8+ cells that completely lost their capacity to release IFN-γ 31. Finally, IL-17-producing CD8+cells have been shown to mediate CD8+ T–cell-mediated autoimmune colitis 32.

Nevertheless, the specific in vitro conditions for differentiation of such cells (that in analogy to their CD4+ counterparts should be termed Tc17 cells) as well as their developmental pathways and molecular requirements have not yet been resolved. In the current study, we found that CD8+ T cells can be stimulated in vitro to polarize into an IL-17-secreting effector population. As described for CD4+ T cells 10–14, the cytokines IL-6 or IL-21 along with TGF-β determine this pattern also in CD8+ T cells. The IL-17-producing CD8+ T effector cells differentiate along a developmental program that is very similar to the one described for Th17 cells 11, 15, 16 which includes the cytokines IL-17 and IL-21, the expression of IL-23R and the lineage-specific transcription factors RORα and RORγt. However, the Tc17 lineage is distinguished from CTL not only by the expression of a type 17 phenotype, but also by the drastically reduced expression of the hallmarks that are central to CTL phenotype and function. Thus, Tc17 cells express greatly diminished levels of the cytotoxic molecule granzyme B, the cytokine IFN-γ and the transcription factor Eomes which is critical for lytic effector differentiation of CD8+ T cells 4. Consequently, these cells exert greatly impaired cytotoxic function, which is not caused by impaired proliferation during development as described for Runx3-deficient cells 21. These data suggest a dichotomy in the generation and function of CTL and Tc17 cells in vitro. Moreover, we provide data describing the presence of Tc17 cells during EAE that are negative for granzyme B and in a majority also for IFN-γ (Fig. 4). This suggests that the described dichotomy in the generation and function of CTL and Tc17 cells occurs also in vivo. However, closer characterization of the function of Tc17 cells in vivo needs to be established.

Mechanistically, the generation of Tc17 cells, similarly to Th17 cells 11, 26, 27, is at least partially dependent on STAT3. After transient silencing of STAT3 in CD8+ T cells using siRNA, the capacity of IL-6 or IL-21 along with TGF-β to induce IL-17 production was markedly impaired as compared with cells treated with controlled Scr (Fig. 3B and C). Moreover, the level of the molecules specific for Tc17 differentiation was downregulated (Fig. 3B and C). In contrast, expression of the CTL-hallmark molecules IFN-γ and granzyme B increased and consequently, mRNA analysis revealed the upregulation of Eomes after transient STAT3 knock-down. Thus, STAT3 seems to be involved in a reciprocal regulation of Tc17 and CTL differentiation.

RORγt is the key transcription factor during differentiation of Th17 cells 11, 16. As the mRNA analysis revealed a correlation between RORγt expression and Tc17 phenotype, we evaluated the causality between RORγt and induction of IL-17 in CD8+ T cells. Notably, under CTL conditions the forced expression of RORγt was able to induce IL-17 production at very low level (Fig. 2C). Possibly, in CD8+ T cells the high level of the transcription factors Eomes and T-bet prevents RORγt function. Indeed, under conditions of reduced Eomes expression, namely in cells cultured with IL-6 or IL-21 alone or along with TGF-β, RORγt was sufficient to induce or to upregulate IL-17 production (Fig. 2C). These results are consistent with the finding that CD8+ T cells doubly deficient for Eomes and T-bet express RORγt at a high level and spontaneously develop a Tc17 phenotype 8. Nonetheless, RORγt seems to be a critical transcription factor in directing IL-17 production in CD8+ T cells.

While our manuscript was prepared, a novel study was published that also analyzes conditions for Tc17 generation in vitro and uses an antigen in the presence of APC to derive such cells 29. Also, the appearance of Tc17 cells in vivo during an infection with influenza virus is demonstrated. Our study confirms and extends the other study by inducing and studying Tc17 cells in an APC-free system, by analyzing the transcription factors Eomes and RORα by proving the importance of STAT3 and RORγt during the differentiation process and by demonstrating the presence of Tc17 cells in the CNS during an autoimmune disease, EAE.

Collectively, our data expand the knowledge about CD8+ T-cell subsets and probably have important implications for immune regulation in the context of pathogenic and protective immunity. Although the pathogenic potential of Tc17 effectors has been emphasized in numerous studies 30–32, the function of these cells during host defenses remains to be defined in more detail. Given potent IL-17 production on the one hand and impaired cytotoxic activity as well as IFN-γ production on the other, it is likely that Tc17 cells have evolved to handle with pathogens distinct from those for which CTL lineages are necessary. Although this idea is speculative, it is possible that Tc17 cells that arise under inflammatory conditions like their Th17 counterparts support them to eradicate extracellular pathogens. Proof of this hypothesis will require further studies in the future.

Materials and methods

CD8+ T-cell purification, in vitro stimulation and staining

WT C57BL/6 mice (purchased from the Jackson Laboratory) and OT1 mice 33 on the C57BL/6 background were bred at the animal facility of the Biomedical Research Center, University of Marburg. CD8+ T cells were prepared by magnetic cell sorting (MACS, Miltenyi, Bergisch-Gladbach, Germany) from spleens and LN of 8–12 wk mice and were primed with plate-bound anti-CD3 (5 μg/mL; clone 145-2C11) and soluble anti-CD28 mAb (1.5 μg/mL; clone 37.51) in the presence of recombinant human (rh) IL-2 (50 U/mL; Novartis), anti-IL-4 (10% culture supernatant of clone 11B11) and anti-IFN-γ (5 μg/mL, clone XMG1.2) (“CTL conditions”). Some cultures also received 2 ng/mL rhTGF-β1 (R&D Systems), 50 ng/mL recombinant murine (rm) IL-6 (Peprotech), 50 ng/mL rmIL-21 (R&D Systems), or combinations of these stimuli. Seventy-two hours later (unless otherwise indicated), cells were washed and restimulated with 50 ng/mL PMA and 750 ng/mL ionomycin in the presence of 10 μg/mL brefeldin A (all from Sigma) for 4 h, after which IL-17-, IFN-γ- and granzyme B-producing cells (anti-IL-17 clone eBio17B7, FITC, PE; anti-IFN-γ clone XMG1.2, FITC; anti-granzymeB clone 16G6, PE; all from eBiosciences) were analyzed by intracellular staining, as described 34. Cells were analyzed with a FACSCalibur and the CellQuest Pro software (BD).

Quantitative real-time PCR

Total RNA was isolated from cell pellets with the High Pure RNA Isolation Kit (Roche). cDNA was synthesized with oligo(dT) primers using the RevertAid™ First Strand cDNA Synthesis Kit (MBI Fermentas) and gene expression was examined with an ABI Prism 7700 Sequence Detection System (Applied Biosystems) using the qPCR™ Core Kit for Il21, Ifng and Hprt1 (hypoxanthine–guanine phosphoribosyl transferase) or the SYBR green I qPCR™ Core Kit (both from Eurogentec) for Eomes, Il23r, Rora, Rorc and Hprt1. Levels of each gene were normalized to Hprt1 expression using the ΔΔCt-method, with the lowest experimental value set to 1. The primer sets and probes for Eomes, Ifng, Il21, Il23r, Rora, Rorc were previously described 18, 35.

Cytotoxicity assay

EL-4 cells were labeled with 2.5 μM or 10 μM CFSE for 5 min at RT. Those labeled with 2.5 μM CFSE were further pulsed with 10 nM of the peptide SIINFEKL for 2 h at 37°C and used as target cells. Those labeled with 10 μM CFSE were used as unpulsed control cells. OT1 CD8+ T cells were purified and primed by anti-CD3/28, as described above for 96 h. Control and pulsed EL4 target cells were mixed at a 1:1 ratio and were incubated in triplicate with the OT1 CD8+ T cells at different effector:target ratios, as indicated (Fig. 1, Supporting Information Fig. 2). After 6 h, the cells were analyzed by flow cytometry. Percent specific lysis was calculated as (1−% targets/% control cells)×100. As an example, one particular set of original FACS data is shown in Supporting Information Fig. 4.

Nucleofection

WT CD8+ T cells were nucleofected with either STAT3-specific si1, si2 or with Scr constructs as described 36. These siRNA were stabilized as described 37 and prepared by IBA (Göttingen, Germany). Their sequences were previously described 36. After nucleofection, the cells were cultured under the conditions indicated in the experiments.

Immunoblotting

For the STAT1, phospho-STAT3 (pSTAT3) and STAT3 immunoblots, whole cell lysates were prepared and immunoblotting was performed, as described 36. Briefly, proteins were fractionated by SDS-PAGE, transferred to nitrocellulose membrane, immunoblotted with pSTAT3 Tyr705 (9131; Cell Signaling Technology) antibodies, then reprobed with antibodies to total STAT3 (124H6; 9139; Cell Signaling Technology) or STAT1 (9171; Cell Signaling Technology) or β-actin (Sigma-Aldrich).

Retroviral transduction

The retroviral vector pMSCV containing RORγt-IRES-GFP and the empty control vector containing IRES-GFP were a gift from D. R. Littmann (New York University School of Medicine, New York, USA) 16. WT CD8+ T cells were infected by the retroviruses as described previously for CD4+ T cells 18. After infection, the cells were cultured under the conditions indicated in the experiments. On day 5, the cells were restimulated as described above and then analyzed for GFP- and IL-17-expression.

EAE induction and ex vivo flow cytometry

EAE was induced in C57BL/6 mice by subcutaneous injection of 200 μg MOG peptide (AA 37–50 28: VGWYRSPFSRVVHL; Dr. R. Volkmer, Charité, Berlin, Germany) emulsified in CFA (Sigma) together with i.v. administration of 200 ng pertussis toxin (Sigma) on days 0 and 2. Disease severity was assessed daily as described previously 17. The preparation of LN cells and CNS lymphocytes was performed as described 17. Cells were restimulated in vitro with 50 ng/mL PMA and 1 μg/mL ionomycin for 4 h, the latter 2 h in the presence of 5 μg/mL brefeldin A. Viable cells (LIVE/DEAD Fixable Aqua Dead Cell dye, Invitrogen) were analyzed for surface CD8 (clone 53–67; PerCP, BD) and intracellularly stained IL-17A (ebio17B7; AlexaFluor488, eBioscience), IFN-γ (XMG1.2; PE-Cy7, eBioscience), granzyme B (16G6; PE, eBioscience). Data were acquired with an LSRII (BD) and analyzed with FlowJo software (Tree Star).

Acknowledgements

This work was supported by Deutsche Forschungsgemeinschaft (LO 396-1, GRK 767 and SFB/TR22) and Gemeinnützige Hertie-Stiftung (1.01.1/08/003).

Conflict of interest: The authors declare no financial or commercial conflict of interest.

Ancillary