JCI - The type I IFN induction pathway constrains Th17-mediated autoimmune inflammation in mice

TRIF-deficient mice develop severe EAE. To investigate the involvement of the TRIF pathway in CNS autoimmune diseases, we examined the development of EAE in TRIF mutant mice (TRIF^Lps2/Lps2), which carry a loss-of-function mutation in the TRIF gene. We hereafter refer to these mice as TRIF^–/– mice. WT and TRIF^–/– mice were immunized with myelin oligodendrocyte glycoprotein (MOG) peptide emulsified in CFA. The disease progression was monitored by physical examination and assigned a disease score of 0 to 5 based on the severity of EAE. Although both groups of mice eventually developed autoimmune disease, TRIF^–/– mice exhibited significantly more severe EAE than WT control mice. We observed that TRIF^–/– mice experienced earlier onset of EAE and exhibited higher disease incidence during the acute phase (Figure 1, A and B). For example, at day 10 after immunization, a significant proportion of TRIF^–/– mice developed neurological symptoms, whereas WT mice did not exhibit any signs of EAE until later in the disease course. Moreover, after initial symptom onset, WT mice gradually recovered, while TRIF^–/– mice developed progressive disease with higher EAE scores. To examine infiltration of inflammatory cells, histological analysis was performed on spinal cord sections from these 2 groups. As shown by H&E staining (Figure 2), TRIF^–/– mice developed more severe inflammatory lesions associated with an increased number of inflammatory foci and cell infiltration compared with the immunized WT mice.

Infiltration of Th17 cells in the CNS of TRIF-deficient mice. Recent studies have demonstrated that Th17 cells are critical for the development of autoimmune diseases, including EAE (10, 16, 21, 23). To investigate whether increased inflammation was associated with infiltration of Th17 cells in CNS, we performed immunohistochemical analysis of spinal cord from WT and TRIF^–/– mice. Anti-CD4 and anti–IL-17 staining revealed few CD4⁺ or IL-17–positive cells in the CNS from WT mice 10 days after immunization. In contrast, multiple inflammatory foci filled with CD4⁺ and IL-17–positive cells were observed in the spinal cord sections of TRIF^–/– mice (Figure 2). To further confirm the infiltration of IL-17–producing CD4⁺ cells in CNS, we isolated mononuclear cells from CNS tissues of WT and TRIF^–/– mice. Consistent with our immunohistochemical analysis, intracellular cytokine staining revealed a significant increase in the percentage of IL-17–positive cells among the infiltrating CD4⁺ T cells from immunized TRIF^–/– mice as compared with WT mice (Figure 3A).

Enhanced generation of antigen-specific Th17 cells in TRIF-deficient mice. The increased inflammation and infiltration of Th17 cells in CNS suggest that the TRIF pathway in the innate immune system is involved in the regulation of Th17 development during EAE induction. To test this, total splenocytes were isolated from mice at day 7 after immunization and restimulated with MOG peptide ex vivo. As shown in Figure 3B, splenocytes from TRIF^–/– mice produced significant amounts of IL-17. In contrast, splenocytes from WT mice produced very low levels of IL-17 at this early stage of EAE induction. At day 21 after immunization, even though splenocytes from WT and TRIF^–/– mice produced significant amounts of IL-17 when restimulated with antigen, TRIF^–/– splenocytes still produced much higher levels of IL-17 (Figure 3C). We noticed that IFN-γ production was also relatively high in the TRIF-deficient cells (Figure 3D). In addition, increased serum levels of IL-17 were observed in TRIF-deficient mice after MOG/CFA immunization (data not shown).

While TRIF is highly expressed in innate immune cells, including macrophages and DCs, there are no reports about its expression in T cells. To determine whether T cells express TRIF, highly purified CD4⁺ T cells were obtained through magnetic bead separation followed by FACS sorting. RT-PCR and Western blot analysis revealed that doubled-sorted CD4⁺ T cells expressed TRIF at both the mRNA and protein levels, albeit the expression level was lower than that in total splenocytes and BM-derived macrophages (BMMs) (Supplemental Figure 1, A and B; supplemental material available online with this article; doi: 10.1172/JCI33342DS1). To distinguish the role of TRIF in T cells and in APCs, we performed adoptive transfer experiments using encephalitogenic T cells from WT or TRIF-deficient mice. MOG-specific lymphocytes were isolated from WT and TRIF-deficient mice immunized with MOG peptide plus CFA. Encephalitogenic lymphocytes were restimulated in vitro with the antigen and transferred into WT or TRIF-deficient mice. As shown in Figure 4A, when encephalitogenic T cells from WT mice were used, TRIF-deficient recipient mice developed EAE with a more severe phenotype than that in WT recipient mice. On the other hand, when encephalitogenic TRIF^–/– T cells were used as donor cells, WT recipient mice had less severe EAE compared with TRIF^–/– recipient mice (Figure 4B). Accordingly, lymphocytes isolated from TRIF-deficient recipient mice produced more IL-17 proteins when restimulated with MOG peptide (data not shown). These results suggest that TRIF-mediated IFN pathways in non–T cell compartments, most likely macrophages and DCs, may play a significant role in inhibiting autoimmunity during the effector phase of EAE.

TRIF limits Th17 development through induction of antiinflammatory cytokine IL-27. These results indicate that activation of the TRIF pathway in innate immune cells may limit the development of proinflammatory Th17 cells. To test this hypothesis, naive T cells from WT mice were cocultured with BMMs or BM-derived DCs (BMDCs). As shown by intracellular staining in Figure 4C, while both LPS-stimulated WT and mutant macrophages could promote Th17 differentiation, TRIF^–/– and IFNAR^–/– macrophages activated by LPS induced more IL-17-positive CD4⁺ T cells in coculture experiments. Furthermore, IL-17 production by T cells was significantly higher when cocultured with LPS-simulated TRIF^–/– macrophages. In addition to TRIF^–/– macrophages, LPS-stimulated IFNAR^–/– macrophages could significantly enhance IL-17 production (Figure 4D). In the parallel experiments, we found that LPS-stimulated DCs could promote Th17 development, and TRIF- and IFNAR-deficient DCs could induce slightly higher IL-17 production. However, the amount of IL-17 induced by WT and mutant DCs was much lower than that induced by macrophages (Supplemental Figure 2, A and B). Therefore, in our system, it seems that macrophages promoted robust IL-17 production from T cells. In this study, we also examined IL-17 production from purified TRIF-deficient T cells in the coculture system. We found that IL-17 production from TRIF^–/– cells was slightly higher, but the response pattern was similar. Like WT T cells, TRIF^–/– T cells produced more IL-17 protein when cocultured with TRIF^–/– or IFNAR^–/– macrophages stimulated with LPS (Figure 4E). Thus, our data indicate that the type I IFN pathway was still able to antagonize Th17 development in T cells lacking the TRIF molecule. Together, our results suggest that type I IFN induction and signaling events in macrophages are important for negatively regulating Th17 development.

Next, we investigated whether the TRIF signaling pathway in macrophages could limit Th17 development through production of negative regulators. To test this hypothesis, we analyzed a number of cytokines produced by WT versus TRIF-deficient macrophages in response to LPS stimulation (Figure 5A and Supplemental Figure 3). The defect in the TRIF molecule led to upregulation of proinflammatory cytokines in macrophages. However, the most striking observation was that TRIF-deficient macrophages produced much less IL-27 protein than WT cells. In addition, IL-27 production was also significantly reduced in TRIF-deficient DCs (Figure 5B). This result is consistent with a recent study showing that TLR4-induced IL-27 production in DCs critically depends on TRIF signals (45). IL-27, a cytokine produced by innate cells, has previously been shown to inhibit the differentiation of Th17 cells and inflammatory autoimmune diseases (34, 35, 46). Consistent with previous reports, we found that IL-27 was a potent inhibitor of Th17 differentiation induced by TGF-β and IL-6 (Figure 5C). IL-27 also inhibited IL-17 production from antigen-specific T cells restimulated in vitro (Figure 5D). These data raise the possibility that TRIF signaling pathway may negatively regulate Th17 development through induction of antiinflammatory cytokines such as IL-27.

IFN-β can upregulate IL-27 expression in macrophages in an IFNAR-dependent manner. Since TRIF is vital for LPS-induced type I IFN production, we then addressed whether type I IFN was involved in LPS-induced IL-27 production. Remarkably, IL-27 production was significantly reduced in IFNAR-deficient macrophages after LPS stimulation (Figure 5E). We also found that IFNAR-deficient DCs produced decreased levels of IL-27 (Figure 5F). IL-27 induction by the TRIF pathway may occur via a feedback loop involving IFN-α/β. In addition to IL-27, IFN-α/β may affect expression of other cytokines involved in inflammation. For example, previous studies have shown that type I IFNs downregulate IL-12 (47). Therefore, we analyzed effects of type I IFN on the production of several cytokines, including IL-1β, IL-6, IL-10, IL-12, IL-23, and TNF-α, some of which have been shown to affect Th17 differentiation and expansion. Macrophages and DCs from WT or IFNAR mice were stimulated with LPS, and cytokines released into culture supernatants were measured by ELISA. Among the cytokines tested, as with our previous studies, defects in type I IFN signaling pathways resulted in reduced IL-10 production. Moreover, IFNAR-deficient cells had increased production of LPS-induced proinflammatory cytokines such as IL-1, IL-12, and TNF-α. On the other hand, IL-6 levels were almost not affected by IFN-α/β, and IL-23 production only increased slightly in IFNAR-deficient cells (Supplemental Figure 4). Since DCs are also major innate immune cells that provide cytokines and costimulatory molecules to T cells, we further analyzed cytokine production by DCs. Although the amount of each cytokine produced by these 2 different cell types varied, the overall pattern was similar. The most significant difference is that macrophages usually produced a significantly higher level of IL-27 in response to LPS stimulation. Even though IL-27 plays a critical role in limiting Th17 development and autoimmune response, it does not exclude the possibility that other cytokines may participate in generating the phenotype of IFNAR-deficient mice. It is most likely that the inflammatory cytokine milieu with the reduced IL-27 level in IFNAR-deficient mice may lead to increased production of IL-17 and development of autoimmune disease.

Lack of IFNAR signaling in macrophages leads to increased Th17 development. Thus far, our data suggest that type I IFN–induced IL-27 production may represent an important mechanism for the immunoregulatory role of the TRIF pathway.

Next, we tried to determine whether supernatant, termed conditional medium (CM), from macrophages stimulated with IFN-β has an inhibitory effect. As shown in Figure 6A, the addition of CM from IFN-β–treated WT and TRIF-deficient macrophages, which express IFNAR, inhibits IL-17 production from Th17 culture. In contrast, CM from IFN-β–treated IFNAR-deficient macrophages failed to do so. To further confirm that type I IFNs directly mediate IL-27 expression, we determined whether IFN-β per se can stimulate IL-27 production in macrophages. We were surprised to find that WT macrophages treated with IFN-β produced IL-27 in a dose-dependent manner, whereas IFNAR^–/– macrophages failed to do so (Figure 6B). The defect in IL-27 induction in both TRIF- and IFNAR-deficient macrophages uncovers a potential link between the innate cytokine IL-27 and type I IFNs. The ability of exogenous type I IFN to inhibit IL-23–induced IL-17 production has been demonstrated previously (7). In our study, we also found that both exogenous IFN-α and IFN-β can suppress IL-17 production from T cells stimulated with IL-6 and TGF-β (data not shown). Then, the question is whether IFNAR-mediated signaling events in macrophages are required for negatively regulating Th17 development. Interestingly, as shown in Figure 4, E and G, IFNAR^–/– macrophages induced more IL-17 production by T cells than WT macrophages, even though WT and IFNAR^–/– macrophages produced similar levels of type I IFNs. These findings indicate that type I IFN–induced signaling events in macrophages play a critical role in limiting Th17 development.

To determine whether IL-27 directly contributes to the inhibitory effect of the type I IFN pathway, we utilized an anti–IL-27 antibody to block IL-27 activity. Our results show that supernatants from IFN-treated macrophages inhibited the development of Th17 cells in vitro. When anti–IL-27 antibody was added to block the IL-27 activity in the CM from IFN-treated WT macrophages, the IFN-β–mediated inhibitory effect on Th17 cells was reversed, as demonstrated by IL-17 production and intracellular IL-17 staining from WT T cells (Figure 6C and Supplemental Figure 5A). We also got similar results with TRIF-deficient T cells (Supplemental Figure 5B). Finally, we tested to determine whether IL-27 also contributes to IFN-mediated inhibition of encephalitogenic T cells. Lymphocytes isolated from WT mice immunized with MOG/CFA were restimulated with MOG peptides in the presence of CM from IFN-treated macrophages plus anti–IL-27 antibody or control IgG. As shown in Figure 6D, CM from IFN-treated macrophages suppressed IL-17 production from antigen-specific T cells, and anti–IL-27 antibody could inhibit such suppression. Similar results were also obtained when IFNAR-deficient T cells were used (Figure 6E). These results demonstrate that IFN-mediated IL-27 production by macrophages plays a major role in inhibiting Th17 development.

IFNAR^–/– mice develop severe EAE in vivo. Since defects in the IFNAR signaling pathway in macrophages lead to accelerated Th17 development, we hypothesized that mice with defects in this pathway should develop more severe Th17-mediated autoimmune disease. To test this hypothesis, we examined the development of EAE in IFNAR^–/– mice. As expected, IFNAR^–/– mice were highly susceptible to EAE (Figure 7A). Notably, the lack of IFNAR mainly affected the progression of EAE, but not the onset or the incidence of disease. While most of the WT mice recovered after the disease activity peaked around days 17–20, IFNAR^–/– mice developed progressive and chronic EAE akin to what we observed in the TRIF^–/– mice. Accordingly, increased infiltration of mononuclear cells was observed in the CNS of IFNAR^–/– mice immunized with MOG/CFA (Figure 7B). These data are in agreement with prior studies showing that exogenous type I IFNs can be used to suppress EAE in mice and MS in humans. Consistent with our findings in TRIF- and IFNAR-deficient mice, IRF3^–/– (IRF, IFN regulatory factor) mice were also sensitive to EAE induction (data not shown). To determine whether Th17-mediated inflammation accounts for the severity of EAE in IFNAR^–/– mice, we examined Th17 development in IFNAR^–/– mice. Similar to TRIF^–/– mice, ex vivo restimulation of T cells revealed that IFNAR-deficient splenocytes secreted more IL-17 than did WT cells from mice 12 days after MOG/CFA immunization (Figure 7C). These data suggest that the defect in IFN signaling pathway in IFNAR-deficient mice contributes to Th17 development and EAE progression in vivo. Altogether, the results from this study demonstrate that TRIF-dependent type I IFN induction pathway and IFNAR signaling may serve a protective role in limiting Th17-mediated CNS autoimmune inflammation.

To investigate whether IL-27 induction is involved in the inhibitory effects of IFN-α/β on EAE development in vivo, we examined IL-27 production in the CNS from IFNAR^–/– mice with EAE. After onset of clinical disease, the CNS tissues were isolated from WT and IFNAR-deficient mice and homogenized; then the IL-27 protein levels were measured by IL-27p28–specific ELISA. As shown in Figure 7D, expression of IL-27 was upregulated in CNS when mice developed EAE. Notably, IFNAR^–/– mice with EAE had much less IL-27 protein in the CNS compared with WT mice. These results suggest that type I IFN may upregulate the expression of IL-27 in CNS to negatively regulate the progression of autoimmune disease. The increased production of IL-27 may represent an attempt by the immune system to dampen the inflammatory response. These data also suggest that IFN-mediated IL-27 production may be one of the mechanisms by which IFN-β exerts its therapeutic effects in the treatment of EAE and MS.

To determine whether this TLR/IFN/IL-27 regulatory loop is relevant in vivo, we utilized adoptive transfer experiments combined with anti–IL-27 blockade. Spleen and lymph node cells isolated from immunized IFNAR^–/– mice were restimulated in vitro with MOG peptide in the presence of supernatants from IFN-treated macrophages with or without anti–IL-27 antibody. As a positive control, we also added recombinant IL-27 to the ex vivo culture. Exogenous IL-27 inhibited the development of EAE in IFNAR^–/– recipient mice induced by adoptive transfer of antigen-specific IFNAR^–/– T cells (Figure 8A). Interestingly, CM from IFN-β–treated macrophages could ameliorate the adoptive transfer of EAE. Furthermore, the inhibitory effect of CM was significantly reduced when IL-27 activity was blocked. As shown in Figure 8B, MOG-specific T cells cultured with CM in the presence of anti–IL-27 antibody induced more severe EAE compared with T cells cultured with CM plus control IgG. We also noticed that IL-27 blocking was not completed; this might result from the amount or the affinity of antibody used. Alternatively, other pathways may also contribute to the inhibitory effects of type I IFNs. Further studies using IL-27 or IL-27R knockout mice may clarify this issue.

As our results suggest that type I IFN may suppress EAE development via induction of IL-27, we determined whether IL-27 treatment in vivo could reverse the severe EAE phenotype in IFNAR-deficient mice. To test this, mice were administered purified murine IL-27 right after immunization. Consistent with recent publications (34, 48), our results showed that IL-27 could significantly inhibit EAE development in WT mice (Figure 8C). Strikingly, injection of IL-27 also suppressed the severe EAE phenotype in IFNAR-deficient mice compared with PBS controls. IFNAR-deficient mice treated with IL-27 showed much milder symptoms. In addition, splenocytes from IL-27–treated WT and IFNAR-deficient mice produced much less IL-17 when restimulated with antigen (Figure 8, D and F). Together, these results indicate that the type I IFN induction and signaling pathways may serve a protective role in Th17-mediated CNS inflammation via IL-27 induction.