Smad2/3 and IRF4 Play a Cooperative Role in IL-9–Producing T Cell Induction

This article is featured in In This Issue, p.2023

Transforming growth factor-β is predominantly expressed in the immune system and is an important pleiotropic cytokine with potent immunoregulatory properties (1). TGF-β regulates T cells, dendritic cells, and macrophages (2–5), suppresses Th1 and Th2, and induces induced regulatory T cells (iTreg), effector T cells, Th17, and Th9 (4, 6).

Th9 cells are a recently described subset of effector T cells that secrete IL-9. They develop from naive T cells in the presence of TGF-β and IL-4 or from Th2 cells in the presence of TGF-β in vitro (7–9). IL-9 production requires transcription factors including STAT6 (10), GATA3 (8), IFN regulatory factor 4 (IRF4) (11), and PU.1 (12). In addition, the Notch (13) and OX40 signals (14) and the CGRP/cAMP/protein kinase A pathway (15) have been shown to promote Th9 development. These results suggest that various Th2-type transcription factors and additional signals are involved in Th9 development. Although many pathophysiological functions have been attributed to Th9 cells, the role of TGF-β in Th9 development in vivo remains to be determined.

The main TGF-β signaling pathway is the Smad pathway. TGF-β first binds to the TGF-β receptor, which then activates Smad transcription factors including three structurally similar proteins: Smad2, Smad3, and Smad4. Smad2 and Smad3 are phosphorylated and activated by TGF-β receptor and heterodimerize with Smad4. The activated Smad complex translocates into the nucleus and regulates target gene transcription. TGF-β has been shown to induce Foxp3, a master transcription factor for Tregs. We have demonstrated in experiments with Smad2 and Smad3 double-deficient T cells that Smad2 and Smad3 are necessary for Foxp3 induction in TGF-β iTreg differentiation (4) and suppression of IL-2 production (16). Smad2 and Smad3 are redundantly essential for the suppression of inducible NO synthase expression in macrophages (2). However, the nuclear orphan receptor retinoic acid–related orphan receptor γt (RORγt) induction by TGF-β is independent of the Smad signaling pathway (5). Although Elyaman et al. (13) demonstrated that cooperative interaction of the Notch pathway and Smad3 was required for Il9 promoter activation, the precise mechanism and significance of Smad signaling for IL-9 production remains to be clarified.

In this study, using T cell–specific Smad2- and Smad3-deficient mice, we found that Smad2 and Smad3 play an important role in regulating IL-9 levels in OVA-induced allergic asthma, suggesting that the TGF-β–Smad2/3 pathway is crucial for Th9 development in vivo. We also showed that Smad2 and Smad3 were redundantly essential for IL-9 production in CD4⁺ T cells in cooperation with IRF4.

T cell–specific Smad2 conditional knockout (Lckcre-Smad2^f/f) mice and Smad3^−/− mice have been described previously (4). T cells from Lckcre-Smad2^f/f mice are henceforth described as Smad2^del/del T cells. The Irf4 knockout mice from Dr. K. Honma have been described elsewhere (17). Mice were kept in specific pathogen-free facilities at Keio University. All experiments using mice were approved by the Animal Ethics Committee of Keio University and performed according to the Animal Ethics Committee’s guidelines.

The Ab to Flag-tag (M2) was from Sigma-Aldrich. PerCP-Cy5.5–conjugated anti-CD4 mAb (L3T4) and PE-conjugated anti- Thy1.1 mAb (HIS51) were purchased from eBioscience. Allophycocyanin-conjugated anti–IL-9 mAb (RM9A4) was purchased from BioLegend. Anti-IRF4 goat polyclonal Ab (sc-6059) and anti–hemagglutinin-tag rabbit polyclonal Ab (sc-805) were purchased from Santa Cruz Biotechnology. Smad2- and Smad3-specific mAbs were from Cell Signaling Technology (5339 for Smad2 and 9523 for Smad3).

Murine Smad2 and Smad3 cDNAs were cloned as described previously (2). Mouse Smad3 was subcloned into eMIGR1 or MIThyR vectors. Mouse IRF4 was PCR-amplified from the mouse cDNA library and subcloned into eMIGR1 and pCMV-FLAG vectors. The mouse Il9 promoter was PCR-amplified from mouse DNA and subcloned into a pGL3 luciferase vector (18).

Naive CD4⁺ T cells (CD4⁺CD25⁻CD62L^hi) were prepared from the spleen and lymph nodes as described (19). Briefly, the spleen and lymph node cells from 6–8-wk-old mice were depleted of non-CD4⁺ T cells through the application of biotin-conjugated Abs against CD8a, B220, CD25, CD11b, CD11c, CD49b, and TER-119 in combination with streptavidin microbeads using large depletion columns (Miltenyi Biotec). Cells were then incubated with CD62L microbeads (Miltenyi Biotec), and the CD62^hi cells were positively selected using large selection columns (Miltenyi Biotec). The purity of CD4⁺CD25⁻CD62L^hi cells was consistently >95%.

CD4⁺ T cell primary culture conditions in this study were as follows: Th1 conditions: IL-12 (PeproTech; 20 ng/ml) and IL-4 neutralizing Ab (11B11, 5 μg/ml); Th2 conditions: IL-4 (PeproTech; 20 ng/ml) and IFN-γ neutralizing Ab (R4-6A2; 5 μg/ml); Th9 conditions: IL-4 (20 ng/ml), recombinant human TGF-β (1 ng/ml; R&D Systems), and IFN-γ neutralizing Ab (5 μg/ml); and iTreg conditions: recombinant human TGF-β (2 ng/ml), IL-2 (PeproTech; 10 ng/ml), IFN-γ neutralizing Ab (5 μg/ml), and IL-4 neutralizing Ab (5 μg/ml). All cultures were performed in RPMI 1640 (Invitrogen) supplemented with anti-CD28 (57.31; 1 μg/ml) Abs, 10% FBS, 1% penicillin/streptomycin, 100 nM nonessential amino acids, 2 mM glutamine, and 0.05 mM 2-ME on plate-coated anti-CD3e (2C11; 2 μg/ml).

Retroviral transduction was performed as described (20). Briefly, naive T cells were plated and subjected to the Th2-differentiation conditions described above starting on day 0. On day 2, fresh retroviral supernatants were added, and the cells were centrifuged at 2500 rpm for 2 h at 35°C. After spin infection, the cells were cultured in the appropriate Th cell differentiation media and harvested on day 4 for intracellular cytokine staining and real-time PCR analysis.

Total RNA was extracted using RNAiso Plus (Takara Bio) and subjected to reverse transcription using a High Capacity cDNA Synthesis Kit (Applied Biosystems). PCR analysis was performed using an iCycler iQ multicolor real-time PCR detection system (Bio-Rad) and SsoFast EvaGreen Supermix (Bio-Rad). All primer sets yielded a single product of the correct size. Relative expression levels were normalized to Gapdh.

For IL-9 intracellular cytokine staining, cells were stimulated for 5 h in complete medium with PMA (50 ng/ml) and ionomycin (500 ng/ml; both from Sigma-Aldrich) in the presence of brefeldin A (eBioscience). Surface staining was then performed in the presence of Fc-blocking Abs (2.4G2), followed by intracellular staining for anti–IL-9 Ab with the Fixation and Permeabilization kit (eBioscience) according to the manufacturer’s protocol instructions. Data were acquired on a BD FACSCanto II (BD Biosciences) and analyzed using FlowJo software (Tree Star, Ashland, OR).

293T cells were seeded on 24-well plates and transfected with various amounts of expression vector with PEI. After transfection, 293T cells were cultured for 24 h. All cells were harvested with 100 μl lysis buffer. The luciferase activity was assessed using a luciferase substrate kit (Promega), and luciferase activity was determined using a Packard luminometer (Packard Instruments) and normalized to β-galactosidase as an internal control.

Mice were immunized with an i.p. injection of OVA (Sigma-Aldrich) adsorbed with aluminum hydroxide (alum; Pierce) at a dose of 20 μg OVA/2 mg alum on days 0 and 7. On day 14, mice received an inhaled challenge of nebulized OVA (0.5% in PBS) using an ultrasonic nebulizer (Omron) for 20 min daily from days 1–6. Mice were sacrificed using an i.p. injection of pentobarbital 48 h after the final challenge. The trachea was cannulated, and lungs were lavaged four times with 0.5 ml PBS. Cells recovered in bronchoalveolar lavage (BAL) fluid were counted using trypan blue dye exclusion. Eosinophils, neutrophils, T cells, B cells, and mononuclear cells in the BAL fluid were distinguished by cell size and expression of CD3, B220, CCR3, CD11b, and Gr-1 and analyzed by flow cytometry. The serum from treated mice was analyzed for IL-4 and IL-9 expression using an ELISA.

Lungs were fixed in 3.7% paraformaldehyde. Fixed tissues were embedded in paraffin, cut into 6-μm sections, and stained with H&E for cellular infiltration analysis or periodic acid–Schiff (PAS) for goblet cell hyperplasia analysis as described (21).

Primary T cells were fixed with 1% formaldehyde at room temperature for 60 min and then suspended in an SDS lysis buffer. After sonication by the S220 Forcused-ultrasonicator (Covaris), samples were incubated with 2 μg Abs or control IgG for 4 h at 4°C. After the addition of Dynabeads protein G (Life Technologies), the immunoprecipitates were sequentially washed once each with a low-salt buffer, a high-salt buffer, and an LiCl buffer and twice with a TE buffer. The DNA–protein complex was eluted by heating at 65°C overnight. Proteins were then digested with proteinase K. DNA was recovered using the QIAquick PCR Purification Kit (Qiagen) and then subjected to real-time PCR analysis.

The 293T cells that had been transfected with the indicated plasmids were washed once with ice-cold PBS and lysed in 0.3 ml TNE lysis buffer containing 10 mM Tris-HCl (pH 7.6), 150 mM NaCl, 1 mM EDTA, and 1% Nonidet P-40 supplemented with protease inhibitor mixture (Nacalai Tesque) (22). After sonication, cellular debris was removed by centrifugation at 15,000 × g for 10 min. Protein from cell lysates was precipitated with 2 μg Ab and 20 μl Protein G-Sepharose (GE Healthcare) for 4 h at 4°C. The immune complex was washed three times with TNE buffer. For Western blotting, the immunoprecipitates or whole-cell lysates were resolved using SDS-PAGE and transferred to Immobilon-P membranes (Millipore). The membranes were blotted with the indicated Abs, and the bound Abs were visualized using HRP-conjugated Abs against goat, rabbit, or mouse IgG and Chemi-Lumi One L Western-blotting detection reagents (Nacalai Tesque).

The Duolink In Situ PLA Kit (Olink Bioscience) was used for in situ proximity-ligation assays (23). T cells were attached to glass slides with cytospin, fixed with 4% paraformaldehyde, and then permeabilized for 15 min in 0.1% Triton X-100 in TBS (25 mM Tris and 100 mM NaCl [pH 7.4]). The slides were then blocked for 2 h with 0.5% BSA in TBS in a humidified chamber at 37°C and incubated overnight at 4°C with goat polyclonal anti-IRF4 Ab and rabbit anti-Smad2/3 mAb. After washing with TBST, proximity ligation was performed using the Rabbit PLUS and Goat MINUS Duolink In Situ PLA kits according to the manufacturer's protocol (Olink Bioscience). CD4 was stained with FITC-conjugated anti-CD4 Ab to show the cell shape.

A paired, two-tailed Student t test was used. The p values < 0.05 were considered significant. All error bars represent the SD.

Th9 cells have been implicated in allergic inflammation such as asthma. First, we investigated the effect of Smad2/3 deficiency in T cells in an OVA-induced asthma model, in which IL-9 has been shown to play an important role. Because T cell–specific Smad2 and Smad3 double-knockout mice died within 5 wk after birth and had severe inflammation (4, 24), we used Lck-cre Smad2^flox/flox Smad3^+/− (LckcreS2^f/fS3^+/−) mice.

We immunized mice with OVA and alum and then challenged them with nebulized OVA to induce airway inflammation. As shown in Fig. 1A, infiltration of inflammatory cells, especially eosinophils, in BAL fluid was much lower in LckcreS2^f/fS3^+/− mice than in wild-type (WT ) mice. This was confirmed by histological examination (Fig. 1B, H&E staining). We observed a marked reduction of mucus-producing cells in the lung of LckcreS2^f/fS3^+/− mice compared with WT mice (Fig. 1B, PAS staining). These data indicated that Smad2/3 deficiency in T cells attenuated OVA-induced asthma. In the serum from treated mice, the amount of IL-4 was similar, but IL-9 was lower in LckcreS2^f/fS3^+/− mice compared with WT mice (Fig. 1C). We isolated splenocytes from treated mice and stimulated them with OVA in the presence or absence of TGF-β. OVA stimulation induced asthma-related cytokine expression (Il4, Il9, Ifng, Il5, and Il13). Adding TGF-β inhibited Il4, Ifng, Il5, and Il13 expression, but stimulated Il9 expression (Fig. 1D). In LckcreS2^f/fS3^+/− mice, all of the cytokines except for Il4 were lower than in WT mice. In particular, Il9 levels in LckcreS2^f/fS3^+/− splenocytes were much lower than those of WT splenocytes. These data suggest that reduced T cell–derived IL-9 production ameliorated OVA-induced asthma in LckcreS2^f/fS3^+/− mice.

To examine whether Smad2 and Smad3 are involved in Th9 differentiation in vitro, we stimulated naive CD4⁺ T cells in the presence of IL-4 with (Th9 condition) or without TGF-β (Th2 condition). In WT T cells, IL-9 secretion increased dose dependently in response to TGF-β (Fig. 2A–C). In Smad single-deficient, Lck-cre Smad2^flox/flox (S2^del/del), or Smad3^−/− (S3^−/−) T cells, IL-9 production was partially reduced, and in addition, IL-9 expression in S2^del/delS3^+/− T cells was almost completely suppressed (Fig. 2A–C). These data indicate that Smad2 and Smad3 are redundantly essential for Th9 differentiation.

We then compared the time course of IL-9 expressions under Th2 or Th9 conditions (Fig. 2D). The IL-9 expression was observed from 24 to 48 h in WT T cells. S2^del/delS3^+/− T cells produced less IL-9 than WT T cells under Th9 conditions, and IL-4 level in S2^del/delS3^+/− T cells under Th9 conditions was higher than those under Th2 conditions (data not shown). These data suggested that in the absence of Smad2/3, Th differentiation was skewed toward Th2 under Th9 conditions.

To further confirm the relationship between IL-9 production and Smads, Smad3 was overexpressed in T cells using retroviral gene transfer. We could not express high levels of Smad2 by retrovirus gene transfer into T cells. Forced expression of Smad3 increased IL-9 protein and mRNA expression in WT T cells (Fig. 2E, 2F). In S2^del/delS3^+/− T cells, IL-9 production was completely restored by Smad3 overexpression (Fig. 2E, 2F); however, induction of IL-9 was much lower by the DNA binding domain-mutant Smad3 than WT Smad3 (Fig. 2G), indicating that IL-9 production requires the Smad3 DNA binding domain in T cells. Thus, Smad2/3 seems to induce IL-9 by directly activating the Il9 promoter.

To further confirm a direct effect of Smads on the Il9 promoter, we examined histone modifications in the Il9 promoter region (conserved noncoding sequence [CNS] 1) using a chromatin immunoprecipitation (ChIP) assay. The increase in transcription is associated with histone H3 and H4 acetylation (AcH3 and AcH4) and/or H3 lysine 4 trimethylation (H3K4me3), whereas silenced genes possess H3K27me3 and/or H3K9me3. In WT T cells, compared with Th2 (+ IL-4) and iTreg (+ TGF-β) conditions, AcH4, H3K9me3, and H3K27me3 were similar under Th9 (+ IL-4 + TGF-β) conditions (Fig. 3A). In contrast, AcH3 and H3K4me3 were higher in the Il9 promoter region under Th9 conditions than in that under Th2 and iTreg conditions (Fig. 3A). In S2^del/delS3^+/− Th9 cells, gene silencing-related modifications (H3K9me3 and H3K27me3) were not changed, but gene activation-related modifications (AcH3, AcH4, and H3K4me3) were drastically impaired (Fig. 3B). These data indicate that Smad2/3 promoted activation-related histone modifications in the Il9 promoter rather than affecting gene-silencing histone modification.

We then investigated the mechanism of how Smad2/3 induces Il9 promoter activation. First, we examined the effect of Smad2/3 deficiency on the expression of the Th cell master regulators (Fig. 4A). Th2 key transcription factor GATA3 has been implicated in IL-9 expression (8, 10). Our expression profiles for WT T cells were consistent with the previous report (10). However, Gata3 expression in S2^del/delS3^+/− T cells was similar to that in WT T cells under both Th2 and Th9 conditions, indicating that loss of Th9 induction by Smad2/3 deficiency was not due to reduction in Gata3 expression.

In contrast, Tbx21 (T-bet) and Rorc (RORγt) were higher, whereas Foxp3 was much lower in S2^del/delS3^+/− T cells compared with WT T cells (Fig. 4A). Modified expression of these master genes could be a mechanism for defective Th9 development that results from Smad2/3 deficiency. To investigate this possibility, we used retroviral gene transfer to test whether overexpression of the other master regulators inhibits IL-9 production. GATA3 increased IL-9 production, whereas T-bet, RORγt, and Foxp3 strongly inhibited IL-9 (Fig. 4B). Thus, Smad2/3 deficiency could be the mechanism that induces high Tbx21 and Rorc expression.

It has been known that transcription factor PU.1 (Sfpi1) is essential for Th9 differentiation (12), and Sfpi1 was shown to be a downstream target of TGF-β (10). As shown in Fig. 4A, right panel, Sfpi1 expression was not increased in Th9 compared with Th2 in our experimental conditions. Furthermore, we observed higher levels of Sfpi1 in Smad2^del/delSmad3^+/− T cells in both Th2 and Th9 conditions (Fig. 4A). These data indicate that PU.1 is not the downstream of Smad2/3 in our Th9 conditions. In addition, overexpression of PU.1 did not increase IL-9⁺ CD4⁺ T cells under Th9 conditions (Fig. 4B). These data suggest that PU.1 is not involved in Smad2/3-mediated IL-9 expression at least in our Th9 differentiation condition.

To avoid the effect of such abnormal skewing to Th1 or Th17 by Smad2/3 deficiency, naive T cells were first differentiated into Th2 cells, and then Th9 cells were induced by culturing Th2 cells with TGF-β (9). Similar to normal Th9 differentiation conditions, WT Th2 cells were converted into Th9 cells by TGF-β, whereas S2^del/delS3^+/− Th2 cells expressed little IL-9 using this method (Fig. 4C, left panel). Interestingly, IL-4 expression was suppressed in both WT and S2^del/delS3^+/− T cells (Fig. 4C, right panel). The expressions of Tbx21, Gata3, and Rorc were similar between WT and S2^del/delS3^+/− T cells under this condition (+ TGF-β in Fig. 4D). Therefore, differences in the expression of master regulators between WT and S2^del/delS3^+/− T cells cannot fully explain reduced Th9 differentiation in Smad2/3-deficient T cells.

Because TCR stimulation in the presence of TGF-β alone induces iTregs, but not Th9, additional factors that are induced by IL-4 or TCR signals should cooperatively upregulate the Il9 promoter with Smad2/3. To identify such factors, we constructed an Il9 promoter luciferase vector and screened various genes, such as transcription factors, transcriptional coactivators, and modifying enzymes expressed highly in T cells. Among >300 genes, the transcription factor IRF4 upregulated Il9 promoter activity most efficiently in 293T cells (Fig. 5A). Moreover, Smad2 and Smad3 coexpression enhanced IRF4-mediated Il9 promoter activity, although Smad2 or Smad3 alone did not activate Il9 promoter activity (Fig. 5B). It has been shown that IRF4-deficient T cells cannot differentiate into Th9 cells (11). Forced expression of IRF4 enhanced IL-9 production under both Th2 and Th9 conditions (data not shown). Thus, we focused on IRF4.

IRF4 expression was similar among Th cell subsets (Fig. 5C). During Th9 differentiation, IRF4 levels were increased 24 h after stimulation in naive T cells, and there was no difference in IRF4 expression between WT and S2^del/delS3^+/− T cells (Fig. 5C, right panel). These data suggest that IRF4 expression is induced by TCR signals and is not affected by cytokines.

To investigate whether IRF4 and Smad2/3 cooperatively function in primary T cells, we introduced Thy1.1-marked Smad3 and EGFP-marked IRF4 into Th2 cells using retroviral gene transfer. An individual introduction of Smad3 or IRF4 increased Th9 cells, whereas the introduction of both molecules together further increased Th9 cells (Fig. 5D). These data suggest that Smad2/3 and IRF4 cooperate to induce Th9 development.

To examine whether Smad2/3 directly interacts with IRF4, we expressed Smad2/3 and IRF4 in 293T cells. Smad2 and Smad3 were coimmunoprecipitated with IRF4 (Fig. 6A). Smad2 and Smad3 contain two conserved structural domains in the N terminus and C terminus, respectively, and there are linker sequences between the MH1 and MH2 domains. The MH1 and MH2 domains participate in DNA binding and transcriptional activation, respectively. Results of an immunoprecipitation assay using truncated Smad3 indicated that the MH2 domain of Smad3 interacts with IRF4 (Fig. 6B).

We confirmed that Smad2/3 and IRF4 were colocalized in the nucleus of Th9 cells (Fig. 6C). However, it was difficult to conclude colocalization by conventional fluorescence microscopy, because a substantial amount of Smad2/3 was already localized in the nucleus without exogenous TGF-β (Fig. 6C, Th2). To demonstrate Smad–IRF4 interaction in primary T cells, we used a proximity-ligation assay that can detect two proteins within 40 nm (23), because we barely obtained a sufficient amount of Th9 cells for immunoprecipitation. This assay confirmed a close association between IRF4 and Smad2/3 in Th9 cells compared with Th2 cells (Fig. 6D).

We investigated the functional interaction between IRF4 and Smad2/3 in the Il9 promoter. There are three CNSs in the Il9 regulating region: CNS0, CNS1 (which functions as a promoter), and CNS2 (Fig. 7A, bottom panel). Both IRF4 and Smad2/3 bound to the same Il9 CNS regions in CNS1, CNS0, and CNS2 (Fig. 7A). IRF4 was bound to the Il9 promoter region even in Th0 and Th2 cells, and IRF4 binding was enhanced by Th9 conditions (Fig. 7A, 7B). Conversely, Smad2 and Smad3 recruitment only occurs in the presence of TGF-β. Therefore, these data suggest that Smad2/3 and IRF4 interaction enhanced recruitment of these two factors into the Il9 CNS regions.

IL-9 is also produced by Th17 cells (25, 26). We examined the role of Smad2/3 and IRF4 under Th17 conditions. First, we examined the expression of IL-9 in Th17 conditions. IL-9 was apparently expressed under Th17 conditions, although Il9 levels were much lower than those under Th9 conditions (Fig. 7B). Then we examined recruitment of Smad2, Smad3, and IRF4 to the Il9 promoter by ChIP assay. We observed a significant recruitment of Samd2/3 and IRF4 to the Il9 promoter under Th17 conditions (Fig. 7B). However, recruitment of Smad2 and IRF4 under Th17 conditions was significantly lower than that under Th9 conditions. Smad3 recruitment was also lower under Th17 than under Th9, although it was not statistically significant due to low levels of expression of Smad3. Thus, IL-9 levels were well correlated with the recruitment levels of Smad2/3 and IRF4 to the Il9 promoter, which supports cooperation of Smad2/3 and IRF4 for IL-9 induction.

To confirm this hypothesis, we used S2^del/delS3^+/− and IRF4^−/− T cells. As shown in Fig. 7C and 7D, IRF4 binding to the Il9 promoter region was reduced in S2^del/delS3^+/− Th9 cells under Th9 conditions, but not under Th2 conditions (Fig. 7C). Smad2 and Smad3 recruitments to the Il9 promoter CNS1 region were completely abolished in IRF4^−/− Th9 cells (Fig. 7D). These data indicate that binding of IRF4 to the Il9 promoter region was enhanced by the presence of Smad2/3, whereas Smad2/3 binding to the promoter was completely dependent on IRF4.

In addition, we confirmed the functional cooperation between IRF4 and Smad for IL-9 production. By using retroviral infection, we transduced IRF4 and Smad3 in S2^del/delS3^+/− and IRF4^−/− Th2 cells (Fig. 7E, 7F). In the Smad2/3-deficient condition, forced expression of IRF4 only slightly increased IL-9 production (Fig. 7E, second bar from left). However, IL-9 was highly expressed with Smad3 coexpression (Fig. 7E, rightmost bar). Similarly, Smad3 overexpression did not enhance IL-9 expression in IRF4^−/− T cells (Fig. 7F, second bar from left), whereas it was enhanced in IRF4^+/+ T cells (see Fig. 2D). Again, coexpression of IRF4 and Smad3 synergistically enhanced IL-9 expression even under Th2 conditions (Fig. 7F, rightmost bar). Taken together with the ChIP assay results, these data indicate that both Smad2/3 recruitment to the Il9 promoter region and its activation are completely dependent on IRF4. Conversely, IRF4 alone can bind the Il9 promoter without Smad2/3, but not activate Il9 transcription. Interaction with Smad2/3 is required for tight binding of IRF4 to the Il9 promoter and its activation (Fig. 8).

In this study, we demonstrated that Smad2 and Smad3 are redundantly essential for Th9 development and that they induce IL-9 expression in collaboration with IRF4. IRF4 can bind to the Il9 promoter without TGF-β signals, but it cannot induce Il9 transcription. After stimulation with TGF-β, Smad2/3 was recruited to the Il9 promoter by interacting with IRF4, and this complex promoted active chromatin modification to initiate Il9 transcription. This model may explain why IRF4 alone cannot activate the Il9 promoter even though IRF4 is essential for Th9 development. PU.1 and GATA3 play essential roles in Il9 promoter activation (8, 12). In addition, the Notch1 intracellular domain together with RBP-Jk have been shown to be necessary for IL-9 expression (13). However, we found that IRF4 and Smad2/3, but not PU.1 and GATA3, directly interacted and cooperatively induced IL-9 expression. Therefore, we propose that Smad2/3 and IRF4 are essential partners for stimulating IL-9 induction (Fig. 8).

IRF4 is a member of the IFN regulatory transcription factor family, for which expression is primarily restricted to lymphoid and myeloid cells. In CD4⁺ T cells, IRF4 expression is induced by TCR stimulation, and it is essential for Th2 and Th17, as well as Th9, development (27). Shindo et al. (28) demonstrated a specific association of IRF4 with c-Rel, which is involved in IL-2 and IL-4 promoter activation. IRF4 activates the Ig L chain genes by binding to a specific DNA sequence in the 3′ enhancer regions, in cooperation with the external transcribed spacer family transcription factor PU.1 (29). IRF4 has also been shown to be important for the induction of IL-17. IRF4 binds to the Il17a transcriptional regulatory region, even under Th0 conditions, and IRF4–B cell–activating transcription factor complexes have been shown to precede the access of Th17-related transcription factors, such as RORγt (30, 31). Similarly, in this study, we demonstrated that IRF4–Smad2/3 interaction in the Il9 CNS regions is important to start Il9 transcription. Thus, IRF4 plays a critical role in the production of various cytokines by cooperating with different interaction partners.

We demonstrated that IRF4 and Smad3 overexpression was sufficient to induce IL-9 under Th2 conditions. However, these two factors did not induce IL-9 under Th0 conditions (without IL-4 and TGF-β), indicating that additional factors are required for full Th9 development. IL-4 activates STAT6 and induces GATA3, both of which have been shown to be important for the induction of Th9 (8, 10). STAT6 has been shown to directly bind to the Il9 promoter (10). Thus, IL-4–mediated STAT6 activation and GATA3 expression could be additional events that are required for Th9 development. Further study is necessary to define a complete set of transcription factors for Th9 development. Unlike other Th subsets, a single master transcription factor may not be able to determine Th9.

Elyman et al. (13) discovered RPB-Jκ-binding site (GTGGGA) and Smad3-binding site (GTCTG) in the Il9 promoter. They found that RBP-Jκ and Smad3 binding was detected in Th9 cells but not in the other Th cell types. These consensus sequences were located between CNS1 and CNS0 (Fig. 7A). We observed enhanced recruitments of Smad2 and Smad3 to this region under Th9 conditions compared with under Th2 conditions. Thus, our study is not contradictory to that by Elyman et al. (13). However, our scanning search defined previously unidentified Smad2 and Smad3 recruiting region near CNS1. Furthermore, we observed that both Smad2 and Smad3 recruitments to the Il9 promoter was strictly dependent on IRF4. Thus, Smad-binding sequences on the genomic DNA seem not to be sufficient for Smad2/3 recruitment to the Il9 promoter. Other factors such as Notch and IRF4 are necessary for a transcriptionally active binding of Smad2 and Smad3 to their binding sites on promoter DNA.

Th9 cells have been implicated in airway hypersensitivity (7–9). In humans, Th9 cells are found in the peripheral blood of allergic patients, and this population is rare in nonallergic individuals (6, 32). However, few reports have clearly demonstrated the involvement of Th9 in allergic inflammation. Xiao et al. (14) reported that transgenic expression of OX40 in T cells resulted in deterioration of airway inflammation and enhanced Th9 induction in vivo. Mikami et al. (15) reported that CGRP promoted allergic asthma and Th9 development through the cAMP/protein kinase A pathway. Using RAMP1-deficient mice that specifically lack CGRP signaling, they also showed partial reduction of IL-9 and amelioration of asthma when these mice were immunized and challenged with OVA. These data supported the hypothesis that Th9 is actually involved in airway inflammation. We also showed that Smad2/3-deficient mice had a lower IL-9 level compared with control mice in a model of asthma. These in vivo and in vitro studies strongly support the involvement of Th9 cells in an allergic asthma; however, further study is necessary to clearly demonstrate a direct link between Th9 and allergic airway inflammation. A recent study identified an innate lymphoid cell (innate lymphoid cell 9) that produces IL-9 predominantly in acute airway responses (33). However, it is still unknown whether the TGF-β–Smad2/3 pathway is involved in innate lymphoid cell 9 development as well as Th9 development.

We thank M. Asakawa, N. Shimizu, and S. Tsuruta for technical assistance.

The authors have no financial conflicts of interest.