IL-17-producing CD8+ T cells (Tc17) appear to play a role in a range of conditions, such as autoimmunity and cancer. Thus far, Tc17 cells have been only marginally studied, resulting in a paucity of data on their biology and function. We demonstrate that Tc17 and Th17 cells share similar developmental characteristics, including the previously unknown promoting effect of IL-21 on Tc17 cell differentiation and IL-23-dependent expression of IL-22. Both STAT1 and STAT4 are required for optimal development of Tc17 cells and maximal secretion of cytokines. Tc17 cells are cytotoxic, and they can be either pathogenic or nonpathogenic upon adoptive transfer in the model of autoimmune diabetes. Tc17 cells treated with TGF-β1 plus IL-6 are not diabetogenic, whereas IL-23-treated cells potently induce the disease. IL-17A and IL-17F are necessary but not sufficient for diabetes induction by Tc17 cells. Tc17 cells treated with TGF-β1 plus IL-6 or IL-23 likely differ in pathogenicity due to their disparate capacity to attract other immune cells and initiate inflammation.
The involvement of IL-17-producing CD8+ T cells (Tc17) in various conditions, such as infection, cancer, and autoimmune inflammation, has been documented in both humans and mice; however, unlike Th17 cells, Tc17 cells have received only marginal attention. The basic factors that determine development of Th17 and Tc17 cells appear to be identical. Activation of mouse CD8+ T cells in the presence of TGF-β1 and IL-6 results in expression of retinoic acid-related orphan receptor (ROR)γt4 and development of Tc17 cells (1, 2, 3, 4), analogous to Th17 cells (5, 6). Th17 cells secrete IL-21, which in an autocrine manner promotes their development (7, 8). The presence of IL-21 mRNA in Tc17 cells has been reported recently (9), but the effect of IL-21 on the development of Tc17 cells has not been described. Cytokines known to antagonize development of Th17 cells, IFN-γ (2), IL-2, IL-4, IL-12 (10), and IL-27 (4), appear to have an inhibitory effect on Tc17 cells as well. IL-23 plays a central role in the biology of Th17 cells, while its effect on Tc17 cells has been less well studied. The comparable stimulatory effects of IL-23 on IL-17A secretion by both CD8+ and CD4+ T cells have been observed (10, 11).
Tc17 cells are largely found in lung and digestive mucosa, paralleling distribution of Th17 cells, with Tc17 cells being less abundant (1). Tumor-bearing mice have greater numbers of Th17 and Tc17 cells that are particularly abundant in tumors (2). It appears that tumors, by secreting large quantities of TGF-β1, induce development of IL-17A+ T cells, which in turn promote tumor survival in an IL-17A-dependent manner (12). Tajima et al. demonstrated IL-6-driven spontaneous expansion of Tc17 cells, which induced colitis in their mouse model (13). In multiple sclerosis a high proportion (70–80%) of Tc17 and Th17 cells among brain-infiltrating T cells have been found in active lesions (14). In experimental autoimmune uveitis, immunization with peptide interphotoreceptor retinoid-binding protein (IRBP)1–20 induced autoreactive Tc17 and Th17 cells (15). Mixed populations of these cells induced experimental autoimmune uveitis when transferred into naive mice, but the uveitogenic potential of Tc17 cells alone has not been addressed. Caruso et al. showed that CD8+ and CD4+ T cells in gastric mucosa of helicobacter pylori-infected individuals produce IL-17A (10) in an IL-23- and STAT3-dependent manner. In a contact hypersensitivity model, allergen sensitization induced the development of CD8+ T cell subpopulations that produce IFN-γ or IL-17A (16). In another study, most CD8+ and CD4+ T cell clones derived from lesional psoriatic skin expressed IL-17A, suggesting that skin-infiltrating IL-17A+ T cells contribute to disease pathogenesis (17). Collectively, these reports demonstrate that Tc17 cells play a role in a variety of diseases and homeostatic mechanisms.
Type 1 diabetes (T1D) is caused by the autoimmune destruction of insulin-producing islet β-cells of the pancreas (18). In newly diagnosed patients with T1D, CD8+ T cells represent a significant portion of pancreas infiltrating cells (19), and islet Ag-specific CD8+ T cells are present in their peripheral blood (20). Studies in a NOD mouse model of T1D have indicated that CD8+ T cells inflict damage to islet β-cells both at the early stage in diabetes development and at the final effector phase (21, 22, 23). Other models of diabetes employ mouse strains with transgenic expression of Ags governed by insulin promoter. In RIP-mOVA mice expression of membrane-bound chicken OVA in pancreatic islet β-cells is controlled by rat insulin promoter (24). Transfer of in vitro-activated OT-I CD8+ T cells, specific for the OVA-derived peptide SIINFEKL in the context of H-2Kb, into RIP-mOVA mice induces rapid onset of diabetes (24, 25, 26).
We show herein that Tc17 and Th17 cells are similar in their development, including the previously unknown promoting effect of IL-21 on Tc17 cell development, as well as IL-23-dependent IL-22 expression. Tc17 cells are cytolitic and can have either a pathogenic or nonpathogenic phenotype in vivo, depending on exposure to IL-23. Furthermore, the diabetogenic potential of Tc17 cells is dependent on IL-17A and IL-17F.
Materials and Methods
C57BL/6J, OT-I transgenic (C57BL/6-Tg(TcraTcrb)1100Mjb/J), and RIP-mOVA (C57BL/6-Tg(Ins2-TFRC/OVA)296Wehi/WehiJ) mice and T-bet-, STAT4- and IFN-γ-deficient mice were purchased from The Jackson Laboratory. STAT1-deficient mice on a C57BL/6 background were purchased from Taconic. Experimental procedures were approved by the Institutional Animal Care and Use Committee of Thomas Jefferson University.
Cell preparation and culture
CD4+ or CD8+ T cells were enriched from spleen mononuclear cells by magnetic microbead cell sorting (Miltenyi Biotec). Cells were typically cultured in RPMI 1640 supplemented with 10% FBS (Invitrogen), 2 mM l-glutamine, 1 mM Na pyruvate, 1× nonessential amino acids, 100 μg/ml penicillin, 100 μg/ml streptomycin, and 0.5 μM 2-ME. T cells were stimulated with anti-CD3 (1 μg/ml) and anti-CD28 (1 μg/ml) Abs in 24-well plates (2 ml of media containing 1.5 × 106 cells/well) either in nonpolarizing conditions (no cytokines added) or in Th1 conditions (5 ng/ml IL-12, 5 ng/ml IL-2) or in Th17 conditions (2 ng/ml TGF-β1, 20 ng/ml IL-6). Where indicated in figure legends, cultures were supplemented with anti-mouse IFN-γ (5 μg/ml), anti-mouse IL-4 (5 μg/ml), IL-1β (10 ng/ml), TNF-α (10 ng/ml), or IL-27 (10 ng/ml). CD8+ cells from spleens of OT-I mice were purified using anti-CD8 microbeads (Miltenyi Biotec) and stimulated at a ratio of 1:5 with irradiated (3000 rad) splenocytes and 1 μg/ml SIINFEKL peptide (OVA257–264) in either Tc1 or Tc17 supporting conditions. Seventy-two hours after stimulation, cells were used for flow cytometric analysis or RNA extraction and supernatants were used for cytokine measurement by ELISA.
For all intracellular staining, cells were stimulated for 4 h with PMA (50 ng/ml) and ionomycin (500 ng/ml; Sigma-Aldrich) and treated with GolgiPlug (1 μg per 1 × 106 cells; BD Pharmingen). In the staining procedure, Fc receptors on cells were first blocked with anti-CD16/32 Ab (2.4G2; BD Pharmingen), and surface and intracellular staining with Abs was performed following the manufacturer’s instructions for staining using Fix & Perm reagents (Caltag Laboratories). Data were acquired on a FACSAria (BD Biosciences) and analyzed with FlowJo software (Tree Star).
Total RNA from T cells was isolated by TRIzol extraction (Invitrogen) according to the manufacturer’s directions, and cDNA was synthesized with a reverse transcription kit (Applied Biosystems). Primer pairs for quantitative real-time PCR were ordered from Applied Biosystems. Gene expression was analyzed by TaqMan real-time PCR (Applied Biosystems). Ribosomal 18S RNA was used as an endogenous control in all experiments.
E.G7-OVA (H-2Kb), a derivative of EL4 thymoma transduced to constitutively express OVA, was used as a target in the cytolytic assays. Cytolitic activity was measured using CFSE-labeled target cells (63). Briefly, EL4 and E.G7-OVA cells were labeled with different concentrations of CFSE and mixed in a 1:1 ratio. After 4 h of incubation with activated OT-I cells, 7-aminoactinomycin D, a membrane-impermeable DNA binding dye (BD Biosciences), was added into the CTL reaction to identify dead cells and was then analyzed by flow cytometry. Percentage of Ag-specific cytotoxicity was calculated by the formula: % cytotoxicity = 100 × [1 − (E.G7-OVA/EL4)experimental/(E.G7-OVA/EL4)control].
Splenocytes from OT-I mice (4 × 106/ml) were stimulated with 1 μg/ml SIINFEKL peptide in either Tc1 or Tc17 supporting conditions. After 72 h, CD8+ T cells were purified by magnetic cell sorting (Miltenyi Biotec) and 1.5 × 106 cells were i.p. injected into RIP-mOVA mice. When specified, splenocytes were rested during 2 days in medium containing 2 ng/ml IL-2 and restimulated for an additional 72 h with either the same cytokines used in the first stimulation or with IL-23 (10 ng/ml). RIP-mOVA mice adoptively transferred with OT-I Tc17 cells were treated with 200 μg per injection of neutralizing monoclonal anti-IL-17A Ab (R&D Systems) or 100 μg per injection of polyclonal anti-IL-17F Ab (R&D Systems) at days 2, 4, and 6 after cell transfer. Mice with blood glucose levels >200 mg/dl in 2 consecutive days were considered diabetic.
Isolation of pancreatic infiltrate
Animals were anesthetized and perfused intracardially with ice-cold PBS. Pancreatic tissue was collected, mechanically dissociated, and centrifuged at 400 × g for 10 min at 4°C. The resultant pellet was fractionated on a 60%/30% Percoll gradient by centrifugation at 300 × g for 20 min. Infiltrating mononuclear cells were harvested from the interface and washed twice in 10% RPMI 1640 (Invitrogen).
Purified CD4+ or CD8+ T cells were cultured in Th1 or Th17 supporting conditions as described above. After 60 h, cells were pulsed for 12 h with 1 μCi of [3H]thymidine. Thymidine incorporation was measured using a scintillation counter.
An unpaired, two-tailed Student’s t test was used for statistical analysis. Differences with p values of <0.05 were considered significant.
TGF-β1, IL-6, and IL-21 induce development of Tc17 cells
IL-23 did not induce IL-17A production by activated CD8+ T cells (Fig. 1). TGF-β1 and IL-6 individually were ineffective in inducing Tc17 cells, but together they strongly promoted development of IL-17A+ cells (45.5%). A significant fraction of cells (10.4%) produced both IL-17A and IFN-γ. The combination of TGF-β1 plus IL-21 also induced development of Tc17 cells, although less potently than did TGF-β1 plus IL-6. IL-1β and TNF-α alone did not induce development of Tc17 cells, but they slightly increased the percentage of Tc17 cells in combination with TGF-β1 plus IL-6. IL-2 inhibited development of Tc17 cells. Blocking IFN-γ signaling by Ab (Fig. 1; see Fig. 5,a) and testing of cells deficient in IFN-γ production (see Fig. 3,a) showed that this cytokine also inhibits Tc17 development. Simultaneous neutralization of both IL-2 and IFN-γ had a positive synergistic effect on development of Tc17 cells (Fig. 1; see Fig. 5,a). IL-4, IL-10, and IL-27 also efficiently inhibited Tc17 development induced by TGF-β1 plus IL-6 (data not shown). The above experiments were performed in parallel with purified CD4+ T cells, and in this direct comparison, CD8+ and CD4+ T cells responded similarly to various culturing conditions. OT-I CD8+ T cells activated with Ag, under the same conditions as depicted in Fig. 1, closely paralleled behavior of purified CD8+ T cells activated with Abs (data not shown).
TGF-β1 plus IL-6 induced secretion of IL-17A and IL-17F by Tc17 cells (Fig. 2, a and b). IL-23 significantly increased secretion of IL-17F but not IL-17A when added with TGF-β1 plus IL-6. IL-1β and TNF-α, alone or together, augmented production of both IL-17A and IL-17F to a greater extent than expected from flow cytometry data where they only modestly increased the percentage of Tc17 cells. IL-2 and IFN-γ had a suppressive effect on IL-17A and IL-17F secretion (Fig. 2, a and b, and data not shown). In contrast to CD4+ T cell cultures, we did not detect IL-21 in supernatants of CD8+ T cell culture (data not shown). CD8+ T cells treated with TGF-β1 plus IL-21 also produced IL-17A and IL-17F but less than did TGF-β1 plus IL-6-treated cells. Overall, levels of IL-17F were several times higher than IL-17A (Fig. 2, a and b), and similar data were obtained with OT-I CD8+ T cells (data not shown).
Direct comparison showed similar levels of mRNA for RORγt and IL-23R in Tc17 and Th17 cells (Fig. 2, c and d). The presence of IL-1β in addition to TGF-β1 plus IL-6 did not increase RORγt and IL-23R expression, while IL-23 further up-regulated IL-23R levels. IL-21, in combination with TGF-β1, also up-regulated RORγt and IL-23R, but less potently than did IL-6 (Fig. 2, c and d). Similar to purified cells activated with Abs, Ag-specific activation of OT-I splenocytes in the presence of TGF-β1 plus IL-6 resulted in up-regulation of RORγt and IL-23R in CD8+ T cells. Neutralization of IL-2 and IFN-γ did not lead to higher expression of RORγt, but increased IL-23R levels (supplemental Fig. 3, a and b).5
These results demonstrate a highly similar regulation of Tc17 and Th17 cell development by cytokine milieu.
Effect of T-bet, STAT1, and STAT4 on development of Tc17 cells
T-bet, STAT1, and IFN-γ deficiency resulted in a 2-fold greater percentage of Tc17 cells in the samples treated with TGF-β1 plus IL-6, and in a 4–5 fold with TGF-β1 plus IL-21, compared with wild-type (WT) CD8+ cells (Fig. 3,a and supplemental Fig. 1). In contrast, lack of STAT1 did not increase the percentage of newly developed Th17 cells. STAT4 deficiency did not have an effect on percentage of Tc17 and Th17 cells. IL-17A and IL-17F concentrations in supernatants of T-bet−/− and STAT1−/− CD8+ cells were significantly increased, but they decreased in STAT4−/− cultures when compared with WT CD8+ T cells (Fig. 3, b and c). In contrast, CD4+STAT1−/− T cells produced 3- to 4-fold less IL-17A, but not IL-17F, than did WT culture. STAT4 deficiency in CD4+ cells resulted in lower levels of IL-17A and IL-17F, similar to CD8+ cells. Identical trends, but lower concentrations of IL-17A and IL-17F, were observed when cells were stimulated with TGF-β1 plus IL-21, consistent with a lower percentage of Tc17 and Th17 cells than in TGF-β1 plus IL-6-treated cultures (supplemental Fig. 1).
Effects of IL-23 on Tc17 cells
To explore an effect of IL-23 on differentiated Tc17 cells, we reactivated these cells (second activation) in the presence of IL-23, TGF-β1 plus IL-6, or with no added cytokines. Neither IL-23 nor TGF-β1 plus IL-6 promoted expansion of Tc17 cells in this experimental design (supplemental Fig. 2a). The presence of IL-23 maintained the percentage of Th17 cells during the second activation while addition of TGF-β1 plus IL-6 or absence of exogenous cytokines led to a decrease in their percentage (supplemental Fig. 2a). IL-23, in contrast to TGF-β1 plus IL-6, did not augment proliferation of Tc17 and Th17 cells during the second stimulation (supplemental Fig. 2b).
TGF-β1 plus IL-6 induced IL-17A secretion from Tc17 cells more potently than did IL-23 (Fig. 4,a). In the first activation both CD4+ and CD8+ T cells treated with TGF-β1 plus IL-6 secreted IL-22, but in the second activation IL-23 was required for the secretion of IL-22 (Fig. 4,b). IL-10 was produced by both Th17 and Tc17 cells during the first activation, with higher levels being detected in Th17 cultures (Fig. 4,c). When Th17 and Tc17 cells underwent a second stimulation, TGF-β1 plus IL-6 treatment caused up-regulation of IL-10 while IL-23 did not (Fig. 4 c). Results presented here were obtained with cells that were rested for 48 h without added cytokines between the first and second activation. Inclusion of IL-2 or TGF-β1 plus IL-6 during the resting period did not significantly alter cytokine production during the second activation (data not shown).
Tc17 cells are cytotoxic
OT-I CD8+ T cells readily differentiated into IL-17-producing cells (68%) when cultured with TGF-β1 plus IL-6. Neutralization of IFN-γ, IL-2, or both cytokines increased the yield of Tc17 cells up to 92% (Fig. 5,a). OT-I Tc17 cells were as highly cytolytic against E.G7-OVA targets as Tc1 cells (Fig. 5,b), whereas no lytic activity was detected against EL-4 cells (data not shown). Tc1 and Tc17 cells had equivalent intracellular levels of granzyme B (Fig. 5 c). Activation of OT-I cells induced RORγt expression only in Tc17 cultures (supplemental Fig. 3a), while IL-23R mRNA was strongly up-regulated in Tc17 and to lesser extent in Tc1 polarizing conditions (supplemental Fig. 3b). T-bet mRNA was up-regulated in Tc1 and to a lesser extent in Tc17 cells. When IFN-γ was neutralized in Tc17 cultures, expression of T-bet was inhibited (supplemental Fig. 3c). Perforin mRNA in both Tc1 and Tc17 cells was modestly up-regulated (2- to 3-fold) compared with naive cells (supplemental Fig. 3d).
IL-23 induces pathogenic phenotype of Tc17 cells
OT-I Tc17 cells differentiated or reactivated in the presence of TGF-β1 plus IL-6 were not pathogenic when injected into RIP-mOVA mice (Fig. 6, a and b), while Tc17 cells stimulated with IL-23 induced hyperglycemia. Tc1 cells induced earlier onset of disease than did IL-23-treated Tc17 cells, but subsequently no significant differences were found in blood glucose levels between the two groups (Fig. 6 b).
Analysis of CD45+ cells isolated from pancreases of RIP-mOVA mice that received OT-I cells showed a larger proportion of Gr1+ cells in mice that received IL-23-stimulated Tc17 cells than in those receiving Tc1 cells (19.3% vs 9.1%) (Fig. 6 c). The greatest difference among Gr1+ cells was in the proportion of neutrophils (CD11b+Gr1+ cells (27)) (9.0% for Tc17 group vs 2.3% for Tc1 group). Consistent with the lack of disease in mice transferred with TGF-β1 plus IL-6-treated Tc17 cells, we obtained only a few CD45+ cells from pancreases of these mice. The finding that Tc17 cells attracted a larger proportion of neutrophils in the pancreas than did Tc1 cells is consistent with the well-known effect of IL-17A and IL-17F to induce preferential accumulation of neutrophils at the site of inflammation. Chemotaxis of neutrophils is mediated by tissue-derived chemokines synergistically induced by IL-17A/IL-17F and other cytokines, such as TNF-α. Similar cellular infiltrates enriched in neutrophils have been described in cases where Th17 cells were the principal initiators of the inflammatory process.
In repeat experiments, IL-23-stimulated OT-I Tc17 cells were adoptively transferred into RIP-mOVA mice and neutralizing Abs against IL-17A or IL-17F were injected on days 2, 4, and 6 after cell transfer. Treatment with either Ab prevented disease, while all control mice developed severe hyperglycemia (Fig. 6 d). A small percentage of mice treated with Abs developed mild hyperglycemia ∼1 wk after the last Ab injection, most likely because injected Abs were catabolized by that time.
In an attempt to identify additional factors that mediate the pathogenicity of IL-23-stimulated OT-I Tc17, supernatants of those cells were analyzed with a rodent multianalyte profile for a panel of 58 secreted immune products and compared with nonpathogenic TGF-β1 plus IL-6-restimulated cells. We found higher concentrations of proinflammatory cytokines GM-CSF, M-CSF, and IL-3 when OT-I Tc17 cells were restimulated with IL-23. Additionally, several other cytokines, including IL-1α, IL-1β, and IL-7, and chemokines, including MCP-1, MCP-3, MIP-2, and IP-10, were similar in the two Tc17 populations (supplemental Fig. 4 and data not shown).
Here we demonstrate that development and functional characteristics of Tc17 and Th17 cells are similar in many respects, including two phenotypes with distinct pathogenicity. Tc17 cells can be, analogous to Th17 cells, either pathogenic or nonpathogenic. IL-23 signaling appears to be required for functional maturation of Tc17 cells and acquisition of the full spectrum of effector functions. Surprisingly, robust cytotoxicity and high levels of IL-17A and IL-17F secretion by nonpathogenic Tc17 cells are insufficient for disease induction.
TGF-β1 plus IL-6 induced expression of mRNA for RORγt in CD8+ T cells. Even though neither we nor others demonstrated that RORγt is required for development of Tc17 cells, based on analogy with Th17 cells, we think that this transcription factor directs differentiation of Tc17 cells. Our findings demonstrate that development of Tc17 cells mirrors development of Th17 cells. In most cases we tested both isolated CD8+ T cells and OT-I splenocytes. Both experimental systems gave consistent results, indicating that particular cytokines act directly on CD8+ T cells, and that other immune cells do not significantly modify their effect.
A recent publication reported that T-bet−/−eomesodermin−/− double-knockout Tc17 cells that developed in vivo contain substantial quantities of IL-21 mRNA (9). However, in repeated experiments we did not detect IL-21 in supernatants of Tc17 polarized WT CD8+ T cells and OT-I splenocytes. Perhaps considerable differences between experimental models tested can account for these contradictory findings. Thus, it remains unclear whether Tc17 cells secrete IL-21, thereby influencing their own, as well as Th17, cell development and/or function.
Tc17 cells in the presence of TGF-β1 plus IL-6 secreted more IL-10 than in the presence of IL-23, similar to Th17 cells (28). In contrast to CD4+ T cells, IL-10 has stimulatory effects on CD8+ T cells and induces their recruitment, cytotoxic activity, and proliferation (29). Unlike in experimental autoimmune encephalomyelitis (EAE), where IL-10 has a clearly suppressive role, the effects of IL-10 on diabetes are complex. IL-10 contributes early to the pathology of diabetes in NOD mice in a CD8+ T cell-dependent manner, but later becomes protective against its development (30). Taking into account opposite effects of IL-10 on CD4+ and CD8+ T cells, and its diverse role in EAE and diabetes, it is difficult to predict the significance of IL-10 produced by Tc17 cells in their pathogenicity.
Mediators that strongly activate STAT1, such as IL-27 (4, 31), type I IFNs (32), and IFN-γ (32, 33), efficaciously inhibit development of Th17 cells. We observed that STAT1 antagonizes Tc17 development, whereas it had no effect on yield of Th17 cells. Since development of STAT1−/− and IFN-γ−/− Tc17 cells was enhanced to the same extent, we thought that IFN-γ might be a possible activator of STAT1 in CD8+ T cell cultures. This was confirmed by neutralization of IFN-γ, which sharply increased the percentage of newly differentiated WT Tc17 cells, while having only a minimal effect on STAT1-deficient cells (data not shown). Thus, it seems likely that STAT1 plays the same intrinsic role in Th17 and Tc17 cell development and that observed differences are caused by higher IFN-γ levels in CD8+ vs CD4+ T cell cultures. Surprisingly, STAT1 deficiency resulted in a dramatic decline of IL-17A production by Th17 cells, while production of IL-17F remained unaffected. Weaker staining of Th17 cells for IL-17A corroborated lower level of this cytokine expression in the absence of STAT1. STAT1−/− Tc17 cells also had lower mean fluorescence intensity when stained for IL-17A, contrasting higher levels of IL-17A in their cell culture supernatants. A likely explanation for this discrepancy is that the lower level of IL-17A secretion by STAT1−/− Tc17 cells is compensated by their higher cell frequency compared with WT cultures. The established view that activation of STAT1 is detrimental to development of Th17 cells is based on results obtained with cytokines that led to strong and prolonged activation of STAT1 (31). However, our results indicate that this transcription factor also supports Th17 and Tc17 development and/or function. Likely activators of STAT1 that contribute to development of IL-17-producing T cells are IL-6 (34) and/or IL-21 (35). Our data also indicate that STAT1 has a different role in expression of IL-17A and IL-17F, at least in the case of Th17 cells. STAT1-, STAT4-, and T-bet-deficient T cells polarized with TGF-β1 plus IL-6 and TGF-β1 plus IL-21 exhibited similar trends in development and cytokine secretion, demonstrating that the role of transcription factors tested does not vary with signaling of IL-6 or IL-21.
STAT4 deficiency did not influence percentage of de novo-differentiated Tc17 and Th17 cells. However, STAT4−/− Tc17 cells produced less IL-17A and IL-17F. Cytokine production by Th17 cells was also affected, especially in the case of IL-17A, consistent with observations made by other groups (36, 37). It was proposed that decreased IL-17A production by STAT4−/− Th17 cells stems from impaired IL-23 signaling, as IL-23 activates STAT4 (38), while STAT4 deficiency does not affect cytokine production by TGF-β1 plus IL-6-stimulated cells (36). Since we used purified T cells, which do not produce IL-23, we think that impaired production of IL-17A and IL-17F by STAT4−/− Tc17 and Th17 in the presence of TGF-β1 plus IL-6 is unrelated to IL-23.
T-bet inhibited Tc17 and, to a lesser extent, Th17 cell development, consistent with the reported suppressive effect of T-bet on Th17 cell development (39) (40). It is possible that IFN-γ signaling through activation of STAT1 induced T-bet, which then suppressed development of IL-17A+ cells. Eomesodermin, a paralog of T-bet, is significantly induced in effector CD8+ T cells and plays an important role in IFN-γ production and cytolytic effector mechanisms (41). Recently, Intlekofer and colleagues showed that T-bet and eomesodermin play redundant inhibitory roles in Tc17 development, requiring the absence of both of them for unopposed Tc17 development in vivo (9). However, T-bet appears to contribute to the function of Th17 cells by being necessary for IL-23R expression and full responsiveness to IL-23 (42, 43). Data from Intlekofer et al. (9) do not indicate such a role for T-bet in the biology of Tc17 cells, suggesting a different role for T-bet and/or IL-23R in the biology of Th17 and Tc17 cells.
We found that Tc17 cells express granzyme B and exhibit cytolytic activity similar to Tc1 cells, in contrast with data from Liu et al. (3) showing that Tc17 cells generated in MLR are not cytolytic. We attempted to reproduce experiments of Liu et al., but in our hands after 5 days of MLR, CD4+ cells considerably outnumbered CD8+ cells and polarization of CD8+ T cells with TGF-β1 plus IL-6 was fairly unsuccessful (50.4% of IFN-γ+, 11.3% of IL-17A+, and 18.8% of IL-17A+IFN-γ+ CD8+ cells), thereby preventing meaningful analyses of Tc17 cytotoxicity. In our experiments, Tc17 cells differentiated in the presence of TGF-β1 plus IL-6 were cytolytic, demonstrating that IL-23 signaling is not necessary for this effector function of Tc17 cells.
TGF-β1 plus IL-6-generated Tc17 cells are highly cytotoxic, indicating that this effector function is not the primary determinant of Tc17 cell pathogenicity in vivo. This conclusion is substantiated by the finding that Tc2 cells have much lower diabetogenic potential than do Tc1 cells, despite being as cytolytic as Tc1 cells (44). Direct killing of β-cells by Tc1 effectors is regarded as an important, if not crucial, mechanism of T1D induction by CD8+ T cells (45, 46, 47). In contrast, Ejrnaes et al. found that the diabetogenic potential of Tc1 clones is associated with IP-10 (CXCL10) production but not with cytokine expression, cytolytic activity, or homing characteristics. The IP-10-expressing Tc1 clone led to destructive insulitis characterized by the presence of more T cells and particularly macrophages, suggesting that CD8+ T cells can induce diabetes by recruiting other immune cell types to the site of inflammation (48). We found similar low levels of IP-10 in OT-I cultures treated with TGF-β1 plus IL-6 or IL-23, suggesting that IP-10 does not play a role in diabetogenicity of these cells (supplemental Fig. 4). Nevertheless, we found higher levels of GM-CSF, M-CSF, and IL-3 in Tc17 supernatant treated with IL-23. Murine and human CTLs are known producers of GM-CSF and IL-3 (49, 50). These two cytokines promote the development and maturation of many hematopoietic cells, including dendritic cells, macrophages, granulocytes, and mast cells (51, 52, 53), and they might play an important role in the pathogenicity of IL-23-stimulated Tc17 cells by promoting survival and effector functions of non-T cells at the site of inflammation.
In agreement with conclusions made by other researchers, our data clearly demonstrate that the high cytolytic potential of autoaggressive Tc17 cells is not sufficient for diabetes induction. It is not currently known if cytotoxicity of IL-23-treated Tc17 cells plays a role in disease processes.
Discovery of the Th17 lineage has been intricately linked to IL-23, but to date, understanding of its role in the biology of Th17 cells remains incomplete. Our observation in a Tc17-driven diabetes model parallels observations made in Th17-driven adoptive EAE, where TGF-β1 plus IL-6-treated myelin oligodendrocyte glycoprotein-specific Th17 cells did not induce disease, whereas IL-23-treated cells were highly pathogenic. Differences in survival and in trafficking/homing properties in vivo are not the cause of differential pathogenicity of TGF-β1 plus IL-6- vs IL-23-treated Th17 cells (28). We did not address these aspects in our Tc17-mediated diabetes model, but by analogy with Th17 cells, including robust proliferating properties in vitro, it seems unlikely that survival and trafficking characteristics of TGF-β1 plus IL-6-treated Tc17 cells are the principal reasons for their nonpathogenic nature. IL-23 stimulates a pattern of cytokine/chemokine secretion from Th17 cells that is distinct from the one induced by TGF-β1 plus IL-6 (28). The authors concluded that failure of TGF-β1 plus IL-6-treated Th17 cells to induce EAE is due to diminished secretion of proinflammatory chemokines and inefficient recruitment of inflammatory cells into the CNS (28). It is likely that a difference in secretion of soluble mediators between TGF-β1 plus IL-6- and IL-23-treated Tc17 cells is also the underlying reason for their differential pathogenicity. Supporting this idea, IL-23-treated Tc17 cells attracted sizable inflammatory infiltrate into the pancreas. This infiltrate was enriched in neutrophils, reminiscent of Tc17 cell-mediated wasting syndrome characterized by multiorgan infiltration of neutrophils in T-bet and eomesodermin double-knockout mice (9), and in CNS infiltrate in the Th17 adoptive EAE model (54).
Initial activation of CD8+ T cells in the presence of TGF-β1 plus IL-6 induced secretion of IL-22, but already differentiated Tc17 cells, like Th17 cells (28, 55, 56), required IL-23 for IL-22 production. The IL-22 receptor is expressed in the pancreas (57), allowing the possibility that IL-22 contributes to diabetes induction by these cells. IL-22 does not play a role in EAE (58), demonstrating that factors other than IL-22 determine encephalitogenicity of IL-23-treated Th17 cells. Furthermore, IL-22 can play an antiinflammatory role, as illustrated by its protective effect in T cell-mediated liver inflammation (59).
Neutralization of IL-17A or IL-17F prevented diabetes induction by Tc17 cells. Almost all IL-17A produced by Th17 cells is secreted as a heterodimer with IL-17F (60, 61), and it is expected that Tc17 cells also secrete these cytokines in the form of heterodimers. Thus, Ab against each of these cytokines simultaneously neutralizes both homo- and heterodimers, making it impossible to distinguish their individual roles in disease pathogenesis. TGF-β1 plus IL-6-treated Tc17 cells produce high levels of these cytokines but are not pathogenic, demonstrating that IL-17A and IL-17F are necessary but not sufficient for induction of Tc17-driven diabetes. IL-17A has also been shown to be crucial to diabetes induction in NOD mice by adoptively transferred BDC2.5 Th17 cells (62).
In conclusion, Tc17 cells can differ in their pathogenicity independent of cytotoxicity and IL-17A and IL-17F secretion. The likely reason for differential pathogenicity of Tc17 cells is the difference in their capacity to attract other immune cells due to the distinct pattern of soluble mediators that they secrete. The relevance of our findings extends beyond the pathogenesis of diabetes and can provide the basis for mechanistic insights in a number of diseases. The noninflammatory nature of TGF-β1 plus IL-6 Tc17 cells may begin to answer the unresolved question why overrepresented tumor-infiltrating IL-17A+CD8+ T cells do not contribute to tumor eradication (2).
We are very grateful to S. Radhakrishnan and L. R. Pease for their help with the diabetes model, V. Puskovic for technical assistance, and K. Regan for editing the manuscript.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by grants from the National Multiple Sclerosis Society (to B.C.) and the National Institutes of Health (to A.M.R.).
Abbreviations used in this paper: ROR, retinoic acid-related orphan receptor; EAE, experimental autoimmune encephalomyelitis; T1D, type 1 diabetes; WT, wild type.
The online version of this article contains supplemental material.