c-Maf Regulates IL-10 Expression during Th17 Polarization¹

Due to its ability to inhibit cytokine production in T cells as well as in APCs (1, 2, 3, 4, 5, 6), IL-10 was originally described as cytokine synthesis inhibitory factor (1, 2). IL-10 plays a critical role in suppressing autoimmunity and inflammatory responses, as IL-10- or IL-10R2-deficient mice develop severe colitis (7, 8), and IL-10 suppresses inflammatory bowel disease in vivo and Crohn’s colitis in humans (9, 10). Recent studies demonstrated that IL-10 produced by Th17 cells restrains the pathologic effects of Th17 (11, 12), further expanding the role of IL-10 in limiting inflammatory responses.

Most analyses of transcriptional regulation of IL-10 have come from studies in macrophages and monocytes. IL-10 gene regulation in these cells has been shown to involve Sp1 (13) activated by p38 MAPK (14), Stat3 (15), cAMP (16, 17, 18), and c-Maf (19). In T cells, IL-10 was shown to be regulated by Jun proteins (20), and chromatin structural differences within the IL-10 gene in differentiated Th1 and Th2 cells have been observed so that committed Th1 cells develop repressive histone modifications at the IL-10 promoter, whereas committed Th2 cells develop an open chromatin configuration (21), supporting the notion that IL-10 may be a Th2 cytokine. Recent studies demonstrated that IL-10 regulation in adaptive T cell response is more complex, with multiple Th2-independent pathways being documented. IL-6 combined with TGFβ Th17-polarizing signals induce Stat3-dependent IL-10 production (11). IL-27 promotes Stat1-dependent IL-10 production in both Th1- and Th2-polarizing conditions (11, 22). The Notch pathway mediates Stat4-dependent IL-10 production both in developing and established Th1 cells (23).

c-Maf, the cellular homolog of the avian viral oncogene v-maf, was the first Th2-specific transcriptional factor identified. It belongs to the AP-1 family of basic region/leucine zipper factors and binds to a consensus site (MARE (Maf recognition element)) in the proximal IL-4 promoter, and it directly activates IL-4 gene transcription (24). Forced expression of c-Maf results in IL-4 production, IFN-γ inhibition, and attenuation of Th1 development (25). Although it has been showed that c-Maf deficiency did not affect IL-10 production under Th2 polarization (26), studies in macrophages showed that c-Maf played a critical role in regulating IL-10 production (19). The role of c-Maf in T cell IL-10 regulation under other polarizing conditions has not been examined.

In this study, we took advantage of the characteristic of Stat6 and T-bet double-deficient cells in which IL-6 plus TGFβ differed from IL-6 plus IL-23 for IL-10 production despite similar IL-17 production (27). We identified that c-Maf plays an important regulatory role in IL-10 expression during Th17 polarization. In contrast to IL-4, c-Maf-regulated IL-10 expression was not strictly dependent of the Stat6/GATA-3 Th2 signal cascade. We identified a MARE motif in mouse IL-10 promoter and provided a molecular basis of IL-10 regulation by c-Maf. Our finding that TGFβ is synergic with both IL-6 and IL-27 for c-Maf expression and consequent IL-10 production further supports the important regulatory role for c-Maf in TGFβ-mediated antiinflammatory activities.

BALB/c, C57BL/6, Stat6 knockout (KO)³ (BALB/c background), T-bet KO (BALB/c background), and IFN-γ KO (BALB/c background) mice were purchased from The Jackson Laboratory. Stat1 KO (129Sv/Ev background) and 129Sv/Ev control mice were purchased from Taconic Transgenics. Stat6/T-bet double-deficient (DKO, BALB/c background) and STAT3^−/− (CD4 Cre:STAT3 f/f) cells were generated as previously described (27, 28). Recombinant mouse IL-2, IL-6, IFN-γ, and PE-anti-IL-17A mAbs were purchased from BD Pharmingen. Recombinant TGFβ was purchased from R&D Systems. Recombinant mouse IL-23, IL-27, anti-CD3, anti-CD28, anti-IFN-γ, anti-IL-4, allophycocyanin-anti-CD4, allophycocyanin-anti-IL-10, FITC-anti-IL-10, PE-anti-IL-4, PE-anti-IFN-γ, and FITC-anti-Foxp3 mAbs, intracellular staining kit, and monesin solution were all purchased from eBioscience. CD4, CD8, and CD25 microbeads were purchased from Miltenyi Biotec. Anti-c-Maf was purchased from Santa Cruz Biotechnology. PMA, ionomycin, and mitomycin C were purchased from Sigma-Aldrich.

CD4⁺CD25⁻ and CD8⁺ T cells were selected by using microbeads or flow cytometry sorting and were seeded in 24-well flat bottom plates. For APC-dependent assays, 2 × 10⁵ cells were cultured with mitomycin C-treated T-depleted splenocytes (1 × 10⁶) plus IL-2 (1 ng/ml) and soluble anti-CD3 (1 μg/ml), combined with various cytokines as indicated. For APC-independent assays, 5 × 10⁵ cells were stimulated with plate-bound anti-CD3 (5 μg/ml) and soluble anti-CD28 (1 μg/ml), along with various cytokines. ELISA for IL-10 was performed using the mouse IL-10 ELISA kit (eBioscience) according to the manufacturer’s protocol. Retroviral supernatants were generated by transfecting bicistronic MigR1-GFP or MigR1-c-Maf-GFP into EcoPack2–293 cells (Clontech Laboratories). After stimulating with anti-CD3 (5 μg/ml) and anti-CD28 (1 μg/ml) mAbs in combination with IL-2 (5 ng/ml) for 26–30 h, CD4⁺CD25⁻ or CD8⁺ T cells were infected with retroviral supernatant by spin infection. A second spin infection was performed on the following day. GFP⁺ cells were purified by cell sorting 48 h after the first infection. Purified GFP⁺ cells were rested in complete medium with IL-2 (5 ng/ml) for 2 days, washed, restimulated with plate-bound anti-CD3 (5 μg/ml) and soluble anti-CD28 (1 μg/ml) for 48 h, and intracellular staining and real-time RT-PCR were performed. Intracellular staining for IL-10, IL-4, IFN-γ, IL-17, and Foxp3 were performed according to the manufacturer’s protocols (eBioscience).

Affymetrix mouse genome 430 2.0 arrays were used to probe the global gene expression profiles in the Stat6 and T-bet double-deficient mouse CD4⁺CD25⁻ T cells after Th17 polarization. Forty-eight hours after stimulation, total RNAs were isolated using RNeasy Protect Mini Kit (Qiagen) and gene arrays were performed at the Mount Sinai Microarray Shared Research Facility. Data were normalized and analyzed using GeneChip operating software (Affymetrix) and presented as expression of indicated molecules relative to expression of internal control probe sets. The primers for real-time RT-PCR are: cyclophilin A, AGG GTG GTG ACT TTA CAC GC and ATC CAG CCA TTC AGT CTT GG; c-Maf, AGC AGT TGG TGA CCA TGT CG and TGG AGA TCT CCT GCT TGA GG; IL-10, AAC TGC ACC CAC TTC CCA GTC and CAT TAA GGA GTC GGT TAG CAG; IL-4, GAA GCC CTA CAG ACG AGC TCA and ACA GGA GAA GGG ACG CCAT; IFN-γ, TGG CTC TGC AGG ATT TTC ATG and TCA AGT GGC ATA GAT GTG GAA GAA; GATA-3, GAA GGC ATC CAG ACC CGA AAC and ACC CAT GGC GGT GAC CAT GC; T-bet, CGG AGC GGA CCA ACA GCA TCG TTT C and CAG GGT AGC CAT CCA CGG GCG GGT; IL-17A, CTC CAG AAG GCC CTC AGA CTA and AGC TTT CCC TCC GCA TTG ACA; retinoic acid-related orphan nuclear receptor (ROR)γT, ACC TCT TTT CAC GGG AGG A and TCC CAC ATC TCC CAC ATT G. Reactions were performed with the LightCycler system (Roche) and the SYBR Green PCR kit (Qiagen). All experiments were performed at least three separate times.

Cells were harvested and washed twice with PBS. Cell extracts were isolated with M-PER mammalian protein extraction reagents (Pierce). Protein concentration was determined by a Micro BCA protein assay kit (Pierce). Proteins were separated by SDS-polyacrylamide, transferred to polyvinylidene difluoride membranes, probed with antiphosphorylated Stat3 or anti-Stat3, incubated with HRP-conjugated anti-rabbit IgG, and detected by the ECL system (Amersham).

Purified CD4⁺CD25⁻ T cells were stimulated with anti-CD3 and anti-CD28 mAbs, with or without IL-6 plus TGFβ for 2 days. Cells extracts were isolated and incubated with biotin-labeled probes containing c-Maf-binding sequence element (MARE) from IL-10 or IL-4 promoter. For supershift assays, 1 μl of anti-c-Maf Abs was added to the binding reactions mixture. For competition assays, 200-fold molar excess of unlabeled probe was used. Reactions were electrophoresed, transferred, and detected according to the instructions of LightShift chemiluminescent EMSA kit (Pierce). The probes used are as follows: IL-10 MARE, CTA AGG TGA CTT CCG AGT CAG CAA GAA ATA TCG and CGA TAT TTC TTG CTG ACT CGG AAG TCA CCT TAG; IL-4 MARE, CTC ATT TTC CCT TGG TTT CAG CAA CTT TAA CTC and GTG TTA AAG TTG CTG AAA CCA AGG GAA AAT GAG.

ChIP assays were performed according to the manufacture’s protocol (Upstate Biotechnology). Briefly, cells were treated with anti-CD3 and anti-CD28 mAbs, with or without IL-6 combined with TGFβ, cross-linked with formaldehyde at a final concentration of 1%, lysed, and sonicated to shear DNA. After immunoprecipitation with anti-c-Maf at 4°C overnight, Ab/DNA complexes were eluted and crosslinking was reversed by incubating at 65°C for 4 h in 5 M NaCl, and the DNA was then recovered by QIAquick PCR purification kit (Qiagen). Real-time PCR was then used to quantify the precipitated DNA sequences. All experiments were performed at least tjree separate times. The primers used are as follows: IL-4 MARE, GAA TAA CTG ACA ATC TGG TGT and TAT ATA TAG AGT TAA AGT TGC; IL-10 MARE, CTC TCC TCT GAC CAA CTG CC and TGG GTT GAA CGT CCG ATA TT; IL-10 distant, CAG TCA GGA GAG AGG GCA GTG A and TTT CCA ACA GCA GAA GCA AC.

The wild-type IL-10 promoter (IL-10wt) containing MARE binding motif were amplified from BALB/c mouse genomic DNA by PCR with primers carrying restriction sites for NheI and BglII (5′-TTG CTA GCT CCT AAC TTC TCA TGC TGG GAT-3′ and 5′-CCA GAT CTA AAG AAC TGG TCG GAA TGA ACT-3′) and cloned into the TransLucent gene promoter reporter vector (Panomics) as previously described. The mutated IL-10 promoter (IL-10mt) was generated by replacing the TGCTGA MARE binding site with an EcoRI (GAATTC) site. Transient transfection experiments were performed in 6-well plates using 2 μg each of IL-10wt or IL-10mt vectors, in combination with 2 μg GFP or c-Maf-GFP plasmids. For dose-response experiments, IL-10wt or IL-10mt was used at 2 μg, along with different doses of c-Maf-GFP from 0 to 3 μg. Twenty-four hours after transfection, luciferase activity was measured with the luciferase assay kit (Promega) according to the manufacturer’s protocol.

Differentiation of Th17 cells is driven by IL-6 plus TGFβ. Our recent studies showed that IL-6 combined with IL-23 differed from IL-6 combined with TGFβ for IL-10 production and pathogenic activities in CD4 T cells deficient in Stat6 and T-bet, despite similar IL-17 production (27). We performed gene array analysis on Stat6 and T-bet double-deficient cells. We rationalized that by comparing gene expression in cells treated with IL-6 plus TGFβ vs TGFβ alone, we would be able to identify genes specific for standard Th17 polarization and responsible for both IL-17 and IL-10 expression. By next comparing gene expression in cells treated with IL-6 plus TGFβ vs IL-6 plus IL-23, we could eliminate molecules involved solely in IL-17 regulation and obtain genes specifically responsible for IL-10 regulation (full microarray data sets are available at Gene Expression Omnibus: www.ncbi.nlm.nih.gov/geo, accession no. GSE15029). We focused only on genes that increased at least 4-fold in cells treated with IL-6 combined with TGFβ vs TGFβ alone or vs IL-6 plus IL-23. We reasoned that to have more potential biological relevance, genes needed to be expressed at relatively high levels (minimal signal strength of 500 in the array assay). As shown in Table I, a total of 10 probes met these criteria, and two probes were identical for c-maf, the cellular homolog v-maf. Importantly, IL-10 was also one of the nine selected genes. Among these nine molecules, only c-Maf is a transcription factor and had been shown to play a critical role in transactivating IL-4 expression and Th2 differentiation (24, 25, 26). Since IL-10 was originally considered as a Th2 cytokine, and c-Maf regulated IL-10 expression in macrophages (19), we further examined c-Maf expression in Th17 polarization and its potential role in Th17 IL-10 regulation.

Real-time RT-PCR for c-Maf and IL-10 was performed on these DKO cells to confirm the array result. As shown in Fig. 1,A, IL-6 alone induced some c-Maf expression, consistent with our previous findings (29). TGFβ alone also induced some c-Maf expression. Importantly, the combination of IL-6 and TGFβ induced c-Maf at much higher levels. IL-6 plus TGFβ also induced c-Maf expression in wild-type cells (Fig. 1,A), indicating that induction was not an artifact due to Stat6 and T-bet double deficiency. Real-time RT-PCR also confirmed that IL-6 plus TGFβ induce higher levels of IL-10 (Fig. 1 A).

Since our previous studies showed that sustained Stat3 activation was required for c-Maf induction (29), we examined the role of Stat3 for c-Maf by using Stat3-deficient CD4 T cells. As shown in Fig. 1,B, Stat3 deficiency abolished IL-6-induced c-Maf, but did not affect TGFβ-induced c-Maf. Stat3 deficiency also abolished the synergistic effect on c-Maf induction by IL-6 plus TGFβ, further demonstrating the critical role for Stat3 in c-Maf regulation. In agreement with a previous report (11), Stat3 deficiency completely abolished IL-6 plus TGFβ-induced IL-10 (not shown). Western blotting for Stat3 activation showed that 24 h after stimulation there was still measurable Stat3 activation in the TGFβ plus IL-6 but not in the IL-6 alone group (Fig. 1 C), suggesting that TGFβ up-regulates c-Maf expression by sustaining Stat3 activation.

To examine the relationship of c-Maf expression to IL-10 production, dose-response experiments were performed. Wild-type cells were stimulated with either a fixed concentration of IL-6 (10 ng/ml) and variable concentrations of TGFβ (0–5 ng/ml), or a fixed concentration of TGFβ (5 ng/ml) with variable concentrations of IL-6 (0–10 ng/ml). Real-time RT-PCR for c-Maf and IL-10 were performed 2 days after stimulation, and intracellular staining for IL-17, Foxp3, and ELISA for IL-10 production was performed 3 days after stimulation. As shown in Fig. 2, A and B, when the concentration of IL-6 was fixed, increasing concentrations of TGFβ induced higher levels of c-Maf, IL-10, IL-17, and Foxp3. When the concentration of TGFβ was fixed, higher concentrations of IL-6 induced higher levels of c-Maf, IL-10, and IL-17 and decreasing Foxp3 expression (Fig. 2, C and D). The finding that c-Maf correlated with IL-17 expression suggested that c-Maf might play a role in IL-17 regulation. However, since IL-6 plus IL-23 induced higher levels of IL-17 without increasing c-Maf expression in Stat6 and T-bet double-deficient cells (Fig. 1 A), this indicated that c-Maf did not directly regulate IL-17 expression. c-Maf and IL-10 expression did not correlate with Foxp3, indicating that c-Maf and IL-10 induction by IL-6 plus TGFβ were independent of Foxp3 and regulatory T cell development. Taken together, these results showed that c-Maf correlated directly with IL-10 expression during Th17 polarization.

Since c-Maf was the first Th2-specific transcription factor identified, we next examined if Th1 or Th2 signals affected c-Maf and IL-10 expression during Th17 polarization. Intracellular staining and ELISA for IL-10 showed that IL-10 production occurred together after TCR activation of wild-type and T-bet-deficient, but not Stat6- or Stat6/T-bet-deficient, CD4 T cells (Fig. 3,A), confirming the notion that IL-10 is associated with the Th2 developmental signal cascade. Addition of IL-6 plus TGFβ induced only IL-10 but not IL-4, regardless of whether the cells were deficient in Stat6, T-bet, or both (Fig. 3,A), indicating that IL-10 production during Th17 polarization was independent of Th1 or Th2 pathways. Real-time RT-PCR for c-Maf showed that IL-6 combined with TGFβ induced similarly high levels of c-Maf in wild-type, Stat6, T-bet, or Stat6/T-bet double-deficient CD4 T cells (Fig. 3,B), indicating that c-Maf expression during Th17 polarization was also independent of Th1 or Th2 signals. c-Maf activates IL-4 gene transcription (24); however, there was very little IL-4 expression during Th17 polarization despite high levels of c-Maf (Fig. 3). GATA-3 has been demonstrated to play a master role in controlling Th2 and IL-4 expression (30, 31). There was little GATA-3 expression in cells under Th17-polarizing conditions (Fig. 3 B), suggesting that TGFβ limits the ability of c-Maf to activate IL-4 by suppressing GATA-3 expression. Taken together, these results showed that c-Maf and IL-10 regulation during Th17 polarization differed from IL-4 and IL-10 regulation during Th2 polarization and was independent of the Stat6/GATA-3 Th2 signal cascade.

IL-27 has recently been demonstrated to regulate IL-10 production in both Th1- and Th2-polarizing conditions in a mostly Stat1-dependent fashion (11, 22). We next investigated if c-Maf is associated with IL-27-mediated IL-10 regulation. CD4⁺CD25⁻ T cells from wild-type and Stat1-deficient mice were stimulated with IL-27, IL-27 plus TGFβ, or IL-6 plus TGFβ in combination with anti-IL-4 and anti-IFN-γ mAbs. In agreement with previous reports (11, 22), real-time RT-PCR showed that IL-27 induced IL-10 expression (Fig. 4,A). Stat1 deficiency abolished IL-27-mediated IL-10 expression but had little effect on IL-6 plus TGFβ-mediated IL-10 expression. Surprisingly, and in contrast to IL-27 alone, we found that IL-27 plus TGFβ also induced significant IL-10 expression in Stat1-deficient cells (Fig. 4,A), showing that IL-27 and IL-27 plus TGFβ use different mechanisms to regulate IL-10. Measurements of c-Maf expression showed that IL-27 alone induced some c-Maf expression in a Stat1-dependent fashion, while IL-27 plus TGFβ significantly induced c-Maf expression in a Stat1-independent fashion (Fig. 4,A), suggesting a role for c-Maf in IL-10 expression induced by IL-27 plus TGFβ. Stat1 deficiency also allowed IL-27 plus TGFβ to induce significant IL-17 production and RORγT expression (Fig. 4, B and C), suggesting that Stat1 activation by IL-27 negatively regulates its effects on Th17 polarization. Since IFN-γ activates Stat1, we next examined how IFN-γ may affect IL-10 and IL-17 expression in Th17 polarization. As shown in Fig. 4,D, IFN-γ negatively regulated IL-10 and IL-17 expression during Th17 polarization, but IFN-γ deficiency did not allow IL-27 plus TGFβ to induce IL-17 expression (Fig. 4 D), further showing that Stat1 activation by IL-27 directly negatively regulates IL-10 and IL-17 expression during Th17 polarization. Collectively, these results indicate that c-Maf and IL-10 regulation during Th17 polarization is different than IL-27 alone-mediated IL-10 expression, and c-Maf may also play an important role in IL-27 plus TGFβ-mediated IL-10 regulation.

To further determine that c-Maf regulates IL-10 expression, we used a bicistronic GFP retroviral vector to transduce c-Maf into T cells. After 2 days, cells expressing GFP alone or c-Maf-GFP were purified by flow cytometric sorting, rested for 2 days, and then restimulated with anti-CD3 plus anti-CD28 mAbs for an additional 2 days. Intracellular staining and real-time RT-PCR showed that c-Maf significantly increased both IL-4 and IL-10 expression in wild-type CD4 T cells (Fig. 5, A and B). Stat6 deficiency abolished c-Maf-induced IL-4 expression, confirming that IL-4/Stat6/GATA-3 signals were pivotal in controlling IL-4 expression and Th2 differentiation, and that high levels of c-Maf expression alone were not sufficient to efficiently initiate IL-4 transcription. In contrast, there was low but significant IL-10 induction by c-Maf in Stat6-deficient CD4 T cells, indicating that c-Maf regulated IL-10 expression could occur in an IL-4/Stat6/GATA-3-independent fashion. c-Maf induced high levels of IL-10 but no IL-4 in wild-type and Stat6-deficient CD8 T cells (Fig. 5, A and B), likely due to highly restricted IL-4 promoter accessibility in CD8 T cells that strongly favor type 1 polarization (32, 33). Low levels of GATA-3 expression in Stat6-deficient CD4, wild-type CD8, and Stat6-deficient CD8 T cells further confirmed that IL-10 regulation by c-Maf is less strictly dependent on GATA-3 (Fig. 5 C). Taken together, these data demonstrated that c-Maf directly up-regulates IL-10 expression, independent of the IL-4/Stat6/GATA-3 Th2 development cascade.

Sequence analysis of the mouse IL-10 promoter identified a half MARE TCAGCA binding motif, similar to one present in the mouse IL-4 promoter (24), located ∼500 bp upstream of the transcription initiation site (Fig. 6,A), within a region previously shown to have high degree of histone H3 phosphorylation after stimulation (34). To examine if this MARE motif bound c-Maf, cell lystes were isolated from CD4⁺CD25⁻ T cells treated with anti-CD3 plus anti-CD28 mAbs, with or without IL-6 plus TGFβ. As shown in Fig. 6,B, similar to IL-4 MARE, IL-10 MARE bound lysates from cells activated by IL-6 plus TGFβ but not TCR alone. Anti-c-Maf Abs supershifted the protein/DNA complex, and IL-4 MARE oligonucleotides competed IL-10 MARE binding, while IL-10 MARE competed IL-4 MARE binding. To determine whether c-Maf bound to the IL-10 promoter in vivo following IL-6 plus TGFβ stimulation, wild-type CD4⁺CD25⁻ T cells were stimulated with anti-CD3 and anti-CD28 mAbs, with or without IL-6 plus TGFβ, and 48 h after stimulation a ChIP assay for IL-10 and IL-4 MARE sequences was performed using anti-c-Maf for immunoprecipitation. As shown Fig. 6 C, anti-c-Maf enriched for IL-10 MARE in IL-6 plus TGFβ-treated cells by ∼6-fold compared with TCR activation alone, while there was no enrichment for an IL-10 distant promoter region 1000 bp upstream of MARE. In contrast to IL-10 MARE, there was much less enrichment for IL-4 MARE, demonstrating that IL-6 plus TGFβ induced c-Maf differentially bound to the IL-10 and IL-4 promoters, correlating with differential effects on IL-10 and IL-4 gene expression. Stat6 deficiency did not affect c-Maf binding to IL-10 promoter, further indicating that c-Maf binding to IL-10 promoter could occur independently of IL-4/Stat6/GATA-3 Th2 signal cascades.

To further prove that c-Maf transactivates the IL-10 promoter, 500-bp regions of the IL-10 promoter containing MARE motif with or without specific mutations were cloned into a vector containing a minimal TA promoter and the TATA box from the herpes simplex virus thymidine kinase promoter. HeLa cells were transiently transfected with wild-type and mutated reporter constructs, along with control GFP or c-Maf-GFP expressing vectors. As shown in Fig. 6, D and E, transfection of GFP alone did not transactivate reporter gene expression, while c-Maf-GFP induced a strong reporter gene response, and mutating the MARE binding region abolished the significant response. Taken together, these experiments demonstrate that IL-6/TGFβ induced c-Maf transactivates the IL-10 promoter to regulate IL-10 production during Th17 polarization.

Th17 cells play a central role in T cell-driven autoimmune diseases (35), and whether Th17 cells express IL-10 is crucial in determining their protective or pathological consequences (11, 12). Most studies for IL-10 gene regulation have been on innate immune cells, and little is known about its transcriptional regulation in adaptive T cell responses. We showed there are at least three different signaling pathways that regulate IL-10 production in T cells: a Stat6-dependent Th2 development-coupled pathway, an IL-27/Stat1 pathway, and a TGFβ in combination with IL-6- or IL-27-mediated Stat3/c-Maf pathway. We uncovered a novel role for c-Maf in modulating T effector responses through regulating IL-10 expression, and provided new insights for understanding TGFβ-mediated antiinflammatory responses.

Consistent with the original notion that IL-10 could be a Th2 cytokine, we showed that IL-10 production is associated with IL-4 production in Th2-polarizing conditions in a Stat6-dependent fashion (Fig. 3,A). c-Maf transactivated both IL-10 and IL-4 gene transcription (Fig. 6 and Ref. 24), but these cytokines differed in requirements for Stat6/GATA-3 Th2 developmental signals. IL-4 gene regulation was strictly dependent on GATA-3 and Th2 polarization, and TGFβ signals switched c-Maf from transactivating IL-4 to IL-10 by suppressing GATA-3 (Figs. 3,B and 6,C). It has been reported that chromatin remodeling of the IL-10 promoter is associated with chromatin remodeling of Th2 loci (21). However, the finding here that c-Maf bound to the IL-10 promoter in Stat6-deficient cells (Fig. 6 C) suggests that chromatin remodeling of the IL-10 gene locus is also regulated by other signals, which allows c-Maf to regulate IL-10 in a Stat6/Th2 signal cascade-independent fashion.

The IL-10 MARE is in a region previously shown to have high degree of histone H3 phosphorylation after stimulation (34), supporting its active role in IL-10 gene transcription. Studies on human IL-10 promoter demonstrated a MARE motif located to nucleotides −196/−184 relative to the transcription initiation site (19), indicating that the c-Maf/IL-10 signal pathway is conserved. c-Maf-deficient mice are neonatal lethal (25, 36). We are not able to dissect roles of c-Maf, Stat3, or TGFβ signals in IL-10 regulation during Th17 polarization. Since retroviral-transduced c-Maf readily induced IL-10 expression (Fig. 5), this suggests that Stat3 or TGFβ signals may not be directly involved in IL-10 gene transcription by c-Maf. IFN-γ and Stat1 signals negatively regulated IL-10 (Fig. 4,D), and retroviral c-Maf induced much less IL-10 expression in CD4 T cells deficient in Stat6 (Fig. 5 A), indicating that optimal IL-10 production by c-Maf required suppression of IFN-γ and Th1 signals. Retroviral c-Maf induced higher levels IL-10 expression in wild-type CD8, Stat6-deficient CD8 compared with Stat6-deficient CD4 T cells, indicating differential regulation in CD4 and CD8 T cells. Despite high levels of IFN-γ expression, CD8 T cells rapidly lost responsiveness to IFN-γ (37), and thus the differential IL-10 production by retroviral c-Maf in Stat6-deficient CD4 and CD8 T cells may be due to their differential responses to IFN-γ. TGFβ has been shown to inhibit IFN-γ production and Th1 differentiation while promote IL-10 expression (38, 39), so that it may promote IL-10 production during Th17 polarization both by inducing c-Maf and by suppressing IFN-γ.

IL-27 has recently been demonstrated to regulate IL-10 expression in both Th1- and Th2-polarizing conditions (11, 22), and our data further supported its role in IL-10 regulation. IL-27 alone induced IL-10 expression in a Stat1-dependent fashion, but in combination with TGFβ it was able to regulate IL-10 in a Stat1-independent fashion (Fig. 4). Similar to IL-6, IL-27 activates both Stat3 and Stat1; however, IL-6-mediated Stat3 activation is stronger and has a longer duration (11). Stat1 deficiency enabled IL-27 to act similar to IL-6 so that Stat3 activation in combination with TGFβ signals induced c-Maf, IL-10, and IL-17. IL-27 alone induced some c-Maf expression, but Stat1 deficiency decreased c-Maf expression. It is possible that IL-27 activated Stat1 and Stat3 to form Stat1/Stat3 heterodimers that were capable of inducing c-Maf. Stat1 deficiency may allow IL-27 to form only fewer numbers of Stat3 homodimers, resulting in less c-Maf induction. Since IL-27 promotes IL-10 expression in Th1-polarizing conditions (11, 22), c-Maf is expressed in Th2 but not in Th1 cells (24), and since IFN-γ/Stat1/Th1 antagonizes c-Maf, it is unlikely that c-Maf plays a significant role in IL-27/Stat1-mediated IL-10 regulation.

TGFβ has long been associated with antiinflammatory functions. It induces Foxp3 expression and regulatory T cell development to control immune response (40, 41, 42). Regulatory T cells have been suggested to be one of the main sources for TGFβ in Th17 development, and their TGFβ production is critical for suppressing autoimmune disease development (43). Here, we showed a novel pathway for TGFβ to attenuate inflammatory immunity by modulating IL-6 and IL-27 signals to synergistically induce c-Maf expression and consequently up-regulate IL-10 production. Overexpression of c-Maf suppressed Th1-mediated experimental colitis (44). Our data that c-Maf directs IL-10 regulation during Th17 polarization further suggest that manipulation of c-Maf expression may be therapeutically beneficial for immunotherapy designed to engage the protective functions of IL-10 in inflammatory response.

The authors have no financial conflicts of interest.