|Home | About | Journals | Submit | Contact Us | Français|
In contrast to CD4 T cells, CD8 T cells inherently differentiate into IFN-γ-producing effectors. Accordingly, while generation of IFN-γ-producing Th1 CD4 T cells was profoundly impaired in mice deficient for both type-I IFN and IL-12 signaling in response to infection with Listeria monocytogenes, generation of antigen-specific, IFN-γ-producing CD8 T cells was unimpaired. However, a fraction of these CD8 T cells also produced IL-17 in an IL-23-dependent manner. Furthermore, the addition of IL-23 in vitro was sufficient for some naïve CD8 T cells to differentiate into IFN-γ/IL-17 dual-producing cells and was associated with increased expression of ROR-γt and ROR-α. Addition of IL-6 and TGF-β to IL-23 further augmented ROR-γt and ROR-α expression and suppressed Eomes expression, thereby enhancing IL-17 production by CD8 T cells. A loss of cytotoxic function accompanied the production of IL-17, as the addition of IL-6 and TGF-β resulted in a marked reduction of granzyme B and perforin expression. Thus, CD8 T cells retain sufficient plasticity to respond to environmental cues and can acquire additional effector functions in response to their environmental context.
The adaptive immune system provides protection against a broad and diverse array of potential pathogens by activating an antigen-specific response tailored against the invading pathogen. In response to infection with each specific pathogen, the immune response is intricately regulated, allowing for fine-tuning of the pathogen-specific response that balances efficient pathogen eradication with minimal collateral damage to host tissues. Cytokines produced by antigen-presenting cells (APCs) play important roles in fine-tuning the differentiation program of pathogen-specific T cells . For example, during Listeria monocytogenes infection, the cytokines IL-12 and type-I IFNs each can prime naïve CD4 T cells to differentiate into IFN-γ-producing Th1 cells, while the absence of both of these cytokines permits Th17 CD4 T cell differentiation provided that IL-6 and TGF-β are present [2, 3]. Interestingly, while the presence and/or absence of specific cytokines readily alters the differentiation program for antigen-specific CD4 T cells induced in response to Listeria infection, antigen-specific CD8 T cells maintain IFN-γ production even under conditions where IFN-γ production is repressed in CD4 T cells.
Recently, however, additional plasticity in the differentiation program for pathogen-specific CD8 T cells has been demonstrated. The T-box transcription factors T-bet and Eomes were shown to limit CD8 T cell plasticity in a collaborative manner, because in their absence, infection with LCMV induced the development of virus-specific CD8 T cells that produced IL-17 and had impaired IFN-γ production and cytotoxicity . These studies illuminate CD8 T cell intrinsic pathways that actively suppress IL-17 production in order to promote an IFN-γ-producing cytotoxic effector program. Additionally, CD8 T cells that produce IL-17 have been observed in situ in humans with psoriasis  and multiple sclerosis , and in animal models of these and other disorders [7, 8], as well as in healthy adults , but the external signals that fostered the development of these IL-17-producing CD8 T cells are incompletely understood. In vitro, IL-21, a cytokine produced by Th17 and T follicular helper cells, can prime naïve CD8 T cells to repress production of IFN-γ while maintaining cytotoxicity , and CD8 T cells cultured in cytokine conditions that allow naïve CD4 T cells to differentiate into Th17 effectors, e.g., IL-6 plus TGF-β, can produce substantial amounts of IL-17 and repress IFN-γ production [11, 12]. Additionally, IL-23 or IL-23 plus IL-1 promote IL-17 production by total CD8 T cells in vitro [7, 13]. Here we describe the cell extrinsic factors regulating the differentiation of antigen-specific CD8 T cells into an IL-17/IFN-γ−dual-producing population and demonstrate how the balance of cell intrinsic factors, such as T-bet, Eomes, and RORγt, is altered to achieve this shift in function.
C57BL/6 mice, IL12Rβ2-deficient and IL-12/23p40-deficient mice backcrossed 10 or more times to B6 were obtained from the Jackson Laboratory and bred in house. P14 TCR-Tg mice specific for the LCMV epitope GP33–41 and congenic for Thy1.1 and type-I IFN receptor-deficient (IFNAR−/−) mice were each provided by Dr. Murali-Krishna Kaja (University of Washington). IFNAR−/− × IL12Rβ2−/− and IFNAR−/− × IL12/23p40−/− mice were generated by intercrossing. All mice were housed in a specific pathogen free facility at the University of Washington. All experiments were performed under IACUC approved protocols.
For infection, 106 CFU of the previously described Lm-OVAΔactA strain [2, 14] was grown to early log phase (OD600 0.1) in brain heart infusion medium (BD Biosciences) at 37°C, washed and diluted in 200 µl final volume and injected intravenously.
For in vivo depletion, 1.0 mg of purified anti-mouse IFN-γ (clone XMG1.2) or IgG1 isotype control (Bio-X-cell) was injected i.p. into mice 1 day before infection. The peptides OVA257–264: SIINFEKL and GP33–41: KAVYNFATC were each purchased from United Biochemical Research, Inc. (Seattle, WA). To assess endogenous CD8 T cell responses to Lm-OVAΔactA, splenocytes were harvested on day 7 and stimulated with SIINFEKL peptide (1 µM) for 4.5 hours in the presence of Brefeldin A for intracellular cytokine staining or for 72 hours for assessment of culture supernatants as described . For in vitro polarization cultures, IL12/23p40-deficient antigen-presenting cells (CD4−CD8−NK1.1−CD44lo) and naïve (CD62LhiCD44lo) T cells from P14 Tg mice were purified on a FACS Aria. 2.5×105 T cells and 7.5×105 APCs were plated into 96-well round bottom plates and stimulated with the P14 GP33–41 peptide (300 ng/ml) in Iscove’s modified Dulbecco’s medium (cIMDM) supplemented with 10% FBS (vol/vol), penicillin, streptomycin, gentamycin, and 50 nM 2-mercaptoethanol. Culture conditions, as indicated, included 10 ng/ml recombinant IL-23 (eBioscience), 20 ng/ml recombinant IL-6 (PeproTech, Inc.), 10 ng/ml recombinant IL-12/23p40 homodimer, 5 ng/ml rhTGF-β1, 5 ng/ml of recombinant IL-12, all from R&D Systems, 10 µg/ml anti-mouse/human TGF-β1 (clone 1D11.16.8) and 10 µg/ml anti-mouse IL-6R (clone 15A7), both from Bio-X-Cell. Cells were split on day 2 in fresh cIMDM plus cytokines. Intracellular cytokine staining was performed after 5 days in culture by stimulating cells with phorbol 12-myristate 13-acetate (5 ng/mL) and ionomycin (0.7 µM) for a total of 4.5 hours in the presence of Brefeldin A (Golgi Plug, BD Biosciences). Antibodies for flow cytometry were purchased from either BD Pharmingen or eBioscience, except for anti-human granzyme B-PE that was purchased from Caltag. IL-17 and IFN-γ concentration within the harvested supernatants were analyzed using Luminex Beadlyte Mouse Multi-Cytokine Flex Kit (Upstate) on the Bio-Plex System (Bio-Rad). Alternatively, cultured CD8 T cells (CD8+Thy1.1+) were purified on the FACS Aria, stimulated with PMA/ionomycin for 4.5 hours, and total cellular RNA was extracted (Qiagen). SuperScript II RNase H Reverse Transcriptase (Invitrogen) was used to create cDNA and transcript abundance was analyzed by RT-PCR using TaqMan probes (Applied Biosystems) and the following primer sets: Tbx21 (Mm00450960_m1); Eomes (Mm01351985_m1); Rorc (Mm01261019_g1); Rora (Mm00443103_m1); Ifng (Mm00801778_m1); Il17a (Mm00439619_m1); Il21 (Mm00517640_m1); Il22 (Mm00444241_m1); Gzmb (Mm00442834_m1); Prf1 (Mm00812512_m1); Il12rb1 (Mm00434189_m1); Il12rb2 (Mm00434200_m1); Il23r (Mm00519942_m1); eukaryotic 18s ribosomal RNA.
Differences in supernatant cytokine concentrations, percentage cytokine producing cells and relative transcript levels were evaluated by Student's unpaired t test (Graph Pad, Prism software).
We previously reported that IL-12 and type-I IFNs collaborate to prime the differentiation of antigen-specific IFN-γ-producing Th1 CD4 T cells after Lm-OVAΔactA infection  and that this Th1 response was abolished in IFNARxIL12/23p40 double-knockout (DKO) mice. Despite this defect, these mice mounted a robust, protective IFN-γ-producing CD8 T cell response comparable to that of wildtype mice. Because the IL-12p40 subunit is also shared by IL-23, we sought to explore further the antigen-specific T cell response after Lm infection using IL-12 receptor β2-deficient (IL12Rβ2 KO) mice, which have a more specific defect. Unexpectedly, we found that while the capacity of CD8 T cells from IFNARxIL12Rβ2 DKO mice to produce IFN-γ after in vitro antigen re-stimulation was fully maintained, these cells also secreted substantial amounts of IL-17 (Figure 1A). Consistent with this observation, intracellular cytokine staining revealed a modest, but consistently reproducible population of IFN-γ/IL-17 dual-producing CD8 T cells in addition to cells producing only IFN-γ in the IFNARxIL12Rβ2 DKO mice (Figure 1B, Figure 1C). The IFNARxIL12/23p40 DKO mice had approximately 3-fold fewer IFN-γ/IL-17 dual-producing CD8 T cells than the IFNARxIL12Rβ2 DKO mice (0.04% vs. 0.12%). No IFN-γ/IL-17 dual-producing CD8 T cells were detected in the IFNAR KO and IL-12Rβ2 KO mice using intracellular staining (Figure 1B, Figure 1C), but low amounts of IL-17 were detected in culture supernatants (Figure 1A). By contrast, IL-17 production by CD8 T cells from IL-12/23p40 KO and wildtype mice was not detected by either method. These findings indicate that IL-23 and/or p40 homodimers  support the generation of antigen-specific, IFN-γ/IL-17 dual-producing CD8 T cells when both type-I IFN and IL-12 signals are absent, and that these specific cytokines play important roles in regulating IL-17 production by CD8 T cells.
During Lm infection, type-I IFNs and IL-12 synergize to prime the production of IFN-γ by NK cells . IFN-γ, in turn, antagonizes the development of CD4 Th17 cells both in vitro and in vivo [3, 16–18]. To determine the contribution of IFN-γ to the development of antigen-specific IL-17-producing CD8 T cells that developed in the absence of type-I IFN and IL-12 signaling, we administered IFN-γ neutralizing antibody or IgG1 isotype control antibody one day prior to Lm infection to wildtype and IFNAR-deficient mice.
Administration of anti-IFN-γ to IFNAR KO mice resulted in a 5-fold increase compared to the isotype control in antigen-specific IL-17 production (2.4 ng/ml vs. 0.5 ng/ml) by CD8 T cells (Figure 2A). This increase did not come at the expense of antigen-specific IFN-γ production, which was also increased, but rather paralleled the emergence of an IFN-γ/IL-17 dual-producing CD8 T cell population (Figure 2B, Figure 2C); however, this dual-producing population was less robust than in the IFNAR x IL12Rβ2 DKO mice (Figure 1B, Figure 1C). These results, with those shown in Figure 1, indicate that type-I IFNs, IL-12 and IFN-γ together inhibit the development of IL-17-producing CD8 T cells during Lm infection, and suggest that the increased IL-17 production observed in IFNAR x IL12Rβ2 DKO mice may be due in part to reduced innate immune production of IFN-γ.
To more precisely evaluate the specific cytokine signals required for the generation of IFN-γ/IL-17 dual-producing CD8 T cells, we isolated naïve CD8 and CD4 T cells and stimulated them in vitro with antigenic-peptide presented by APCs from IL-12/23p40 KO mice. These APCs allow the effects of IL-12 and IL-23 on IL-17 and IFN-γ production by T cells to be more precisely addressed.
In the absence of added cytokines, nearly all naïve CD8 T cells differentiated into effectors that produced IFN-γ and none produced detectable IL-17 (Figure 3). When IL-23 alone was added, a small but distinct population of IFN-γ/IL-17 dual-producing CD8 T cells was consistently observed. IL-6 and TGF-β dampened IFN-γ production and induced substantial numbers of IL-17-producing and smaller numbers of IFN-γ/IL-17 dual-producing cells; in these conditions IL-17 production was augmented only slightly by the further addition of IL-23.
Consistent with the inhibitory effect of IL-12 on IL-17 production in vivo, addition of IL-12 abolished the induction by IL-23 of IFN-γ/IL-17 dual-producing CD8 T cells and antagonized the inhibitory effects of IL-6 and TGF-β on IFN-γ production (Figure 3). These findings indicate that in the absence of IL-12, IL-23 supports the differentiation of a fraction of naïve CD8 T cells into effectors that produce both IFN-γ and IL-17 in vitro. This effect was specific to the IL-23 p40/p19 heterodimer, because like IL-12, p40 homodimer induced only IFN-γ-producing cells (Figure 4A). Moreover, increasing the concentration of IL-23 ten-fold augmented the percentage of IL-17-producing cells, while a similar increase in p40 homodimer did not (Figure 4B).
The ability of exogenous IL-23 to support the generation of IFN-γ/ IL-17 dual-producing CD8 T cells in the absence of IL-12 was surprising. To determine whether IL-23 alone was sufficient or if it was acting in concert with IL-6 and TGF-β produced by the cultured cells, we added blocking antibodies to the IL-6R and TGF-β. Addition of these antibodies greatly reduced the numbers of IFN-γ/IL-17 dual-producing CD8 T cells in cells cultured in the presence of exogenous IL-23 or in Tc17 (IL-6, TGF-β, IL-23, anti-IL-4 and anti-IFN-γ) conditions (Figure 4A). These findings indicate that IL-23 is not sufficient to induce IL-17-producing CD8 T cells, but, in the absence of IL-12, can do so in concert with the amounts of IL-6 and TGF-β present in our cultures.
The induction by exogenous IL-23 of IFN-γ/ IL-17 dual-producing CD8 T cells, suggested that these conditions were also sufficient to induce IL-23 receptor expression on these cells. To address this prediction, we assessed expression of mRNA encoding the two chains of the IL-23 receptor. Resting naïve CD8 T cells did not express these receptor chains (Figure 5). IL-12Rβ1 mRNA was induced when cells were activated under all conditions evaluated. In contrast, IL-23R expression was not induced by stimulation with antigen alone, but was markedly upregulated within 48 hours of activation in the presence of exogenous IL-23 or in Tc17 conditions.
T cell effector differentiation is controlled by transcription factors. The transcription factors T-bet and Eomes collaborate to sustain IFN-γ production and cytotoxic function and to repress IL-17 production by CD8 T cells [4, 19, 20]. Both ROR-γt and ROR-α influence the development of CD4 T cells into Th17 cells [21, 22], but their involvement in IL-17 expression by CD8 T cells is less well understood. To explore the role of these transcriptional regulators in the development of IFN-γ/IL-17 dual-producing CD8 T cells, we evaluated their expression under the conditions described above. Expression of Rorc (ROR-γt), Rora (ROR-α), Il17a and Il21 increased modestly when naïve CD8 T cells were activated in the presence of IL-23 alone, more strongly in the presence of IL-6 plus TGF-β and maximally in Tc17 conditions. Reciprocally, IL-23 alone did not substantially alter Eomes or Ifng expression, whereas both were repressed by IL-6 plus TGF-β and in Tc17 conditions. IL-12 blocked this induction of Rorc and Il17a and upregulated Ifng expression, whereas expression of Rora (ROR-α) and Il21 were not affected (Figure 6A,B). Levels of Tbx21 mRNA, the transcript for T-bet, did not vary substantially under these conditions (data not shown). Under none of the polarizing conditions was Il22 mRNA detected. Thus, expression of Il17a by CD8 T cells correlated directly with Rorc expression and inversely with Eomes expression. Conversely, Ifng expression correlated inversely with the expression of Rorc.
A key function of effector CD8 T cells resides in their ability to produce granzyme B and perforin to exert cytotoxic activity on infected cells, the expression of which is upregulated by T-bet and Eomes . Production of IFN-γ and cytotoxic activity in CD8 T cells is not always coordinate, as IL-21 can repress the production of IFN-γ while maintaining cytotoxicity . When we assessed these mediators of cytotoxicity, we found that IL-12 enhanced, TGF-β plus IL-6 and Tc17 conditions inhibited, and exogenous IL-23 alone had no effect on granzyme B abundance (Figure 7A) and granzyme B and perforin mRNA expression (Figure 7B). Together these results indicate that IL-6 and TGF-β are sufficient to dramatically reprogram CD8 T cell function by inducing robust IL-17 production and inhibiting their canonical functions of cytotoxicity and IFN-γ production, whereas IL-23 preserved these canonical functions while inducing low-level IL-17 production.
Taken together, these findings demonstrate that antigen-specific CD8 effector T cells that produce both IL-17 and IFN-γ can arise in response to infection in vivo. These findings corroborate those of Hamada et al , who identified the emergence of an IFN-γ/IL-17 dual-producing CD8 T cell population in the lungs of influenza-infected mice. During Lm infection, IL-17A produced by γδ T cells helps to promote bacterial clearance and the formation of organized granulomas within the liver . Our studies indicate that in the absence of inhibitory signals, IL-17-producing adaptive immune cells can also be generated during Lm infection. We show here that the emergence of these IFN-γ/IL-17 dual-producing CD8 T cells is inhibited by IL-12, type-I IFNs and IFN-γ, which instead promote the canonical cytotoxic, IFN-γ-producing CD8 T cell phenotype. Furthermore, we also show that emergence of these IL-17 producing CD8 T cells in vivo was also dependent on IL-23, because it was markedly reduced in IFNARxp40 DKO mice, which lack IL-23, IL-12, and type-I IFN signaling, while IL-17-producing CD8 T cells were most pronounced in IFNARxIL-12Rβ2 DKO mice, in which IL-23 signaling is retained. When we tested the ability of p40 homodimers to induce IL-17 production in vitro, we found that they were unable to do so, indicating that IL-23 and not p40 homodimers was required.
IL-6 and TGF-β have been clearly implicated in the induction of CD4 Th17 cells, while IL-23 is believed to be more important in the terminal differentiation and survival of Th17 cells [24, 25]. We found that IL-6 and TGF-β were also essential for IL-17 production by CD8 T cells, and previous reports have demonstrated that IL-23 can stimulate the production of IL-17 by memory CD8 T cells [7, 8]. However, to our knowledge, this is the first demonstration that IL-23, when type I promoting cytokines are blocked, is sufficient to promote the development of naïve CD8 T cells into IFN-γ/IL-17 dual-producers. These effects of IL-23 were dependent on endogenous IL-6 and TGF-β and could be amplified by the further addition of exogenous IL-6 and TGF-β. Weaver and colleagues showed that IL-23 could support IL-17 production by CD4 T cells in the absence of exogenous IL-6 and TGF-β when inhibitory signals, such as IL-12 and IFN-γ, were eliminated . Our results show that this is also true for CD8 T cells.
Our report also directly links cell-extrinsic cytokine-triggered IL-17 and IFN-γ production with levels of cell-intrinsic lineage-defining transcription factors. The importance of T-bet and Eomes in CD8 effector function has been appreciated, and gradations of T-bet and Eomes modulate CD8 function and survival [27–29]. Our studies implicate Eomes as an important regulator of CD8 T cell cytotoxic activity, as the repression of Eomes upon the addition of IL-6 and TGF-β paralleled the suppression of granzyme B and perforin. Conversely, inhibition of these cytotoxic proteins and IFN-γ correlated directly with the increased expression of ROR-γt.
In summary, our findings show that depending on the cytokine milieu, CD8 T cells have the ability to differentiate into one of several populations: canonical CD8 effectors producing only IFN-γ, and subsets producing IFN-γ and IL-17, or, at least in vitro, only IL-17. The diversity of populations indicates that although CD8 T cell differentiation is less plastic than for CD4 T cells, CD8 effector function is nonetheless clearly affected by the overall cytokine milieu.
We thank Michael Bevan and Murali Krishna-Kaja for providing TCR transgenic mice.
This work was supported in part by an Infectious Diseases Society of America Career Development Award (S.S.W), a March of Dimes Basil O’Connor Research Award (S.S.W) and National Institutes of Health Grants T32 CA009537 and T32 GM07270 (M.M.C), K08HD51584 (S.S.W.) and R01 HD18184 (C.B.W.).
The authors have no financial conflict of interest.