Identification of a New Pathway for Th1 Cell Development Induced by Cooperative Stimulation with IL-4 and TGF-β

Elucidating the mechanisms by which naive CD4 T cells differentiate into effector T cells is crucial for understanding T cell-dependent immune responses. After AG stimulation, naive CD4 T cells differentiate into several distinct Th cell subsets, such as Th1, Th2, Th9, Th17, and inducible regulatory T cells (iTreg) (1–3). In the presence of TGF-β, IL-12–induced Th1 cells and IL-4–induced Th2 cell differentiation is inhibited, whereas TGF-β supports the development of Th9, Th17, and iTreg. Although TGF-β alone induces iTreg in the presence of IL-2, Th9 cell differentiation is induced by a combination of IL-4 and TGF-β (2), and the presence of IL-6 with TGF-β results in Th17 cell development (4, 5). The master transcription factors that regulate Th1/Th2/Th17/iTreg cell differentiation have been identified. T cell-specific T-box transcription factor (T-bet) appears to be a key factor for Th1 cell differentiation (6, 7) as is Gata3 for Th2 cells (8), Rorα and Rorγt for Th17 cells (9), and Foxp3 for iTreg (10).

IL-12 is an important cytokine that induces Th1 cell development through activation of the STAT4-signaling pathway (11). IFN-γ can also prime Th1 cells independent of IL-12 (12). The activation of Stat1 and the subsequent T-bet induction seem to be essential for IFN-γ–induced Th1 cell development (11). However, in the presence of IL-4, either IL-12 or IFN-γ alone has been reported to be sufficient to induce T-bet and elicits IFN-γ–producing cells (12). In contrast, both IL-12 and IFN-γ were found to be required for effective Th1 cell development if IL-4 is present during the priming phase. These findings suggest the existence of an alternative pathway for Th1 cell development.

IL-4–induced activation of the Stat6-signaling pathway and the subsequent expression of Gata3 are required for Th2 cell differentiation (8, 11). During Th2 cell differentiation, Gata3 binds to various regulatory regions in the Th2 cytokine (IL-4, IL-5, and IL-13) gene loci and induces chromatin remodeling (13–15). IL-4 also blocks Th1 cell development through the downregulation of Th1-related factors, such as Stat4 and IL-12Rβ2 (16). Another mechanism through which Gata3 inhibits Th1 cell development is by blocking the Runx3-mediated pathway. Runx3 was reported to regulate IFN-γ production via binding to the IFN-γ promoter (17). Moreover, the Gata3-mediated inhibition of T-bet functions was reported (18). Therefore, IL-4 represses Th1 cell differentiation and functions through multiple pathways.

TGF-β is a pleiotropic cytokine produced by various cell types, which has multiple effects on the immune response, including proliferation, differentiation, cytokine production, migration, and survival (19, 20). T cells were shown to play essential roles in the severe multiorgan autoimmune inflammatory disease that was observed in TGF-β1–deficient mice (19). TGF-β has been identified as an inducer of regulatory T cells and Th9 and Th17 cells (1–3, 21). Furthermore, TGF-β strongly inhibits Th2 cell development via inhibition of IL-4 signaling and/or Gata3 expression, even in the presence of exogenous IL-4 (22, 23). However, the effects of TGF-β on Th1 cell development are less clear. It is known that TGF-β inhibits IL-12–dependent Th1 cell differentiation, whereas IFN-γ–induced Th1 cell development does not decrease, but rather becomes enhanced, in the presence of TGF-β (24).

In this study, we demonstrate that a combination of IL-4/TGF-β augments the generation of Th1 cells that preferentially express the CD103 molecule if CD4 T cells are primed with IFN-γ. We named the IL-4/TGF-β/IFN-γ–induced Th1 cells “CD103⁺ Th1 cells.” The T-box–containing transcription factor eomesodermin (Eomes) is preferentially expressed in CD103⁺ Th1 cells and regulates IFN-γ production. For CD103⁺ Th1 cell development, the IFN-γ/TGF-β–mediated formation of active chromatin at the Eomes gene locus and subsequent Stat6-dependent transcriptional activation of the Eomes gene seem to be required. Moreover, IFN-γ–producing CD103⁺ Th1 cells are detected in the intraepithelial lymphocytes (IEL) of normal mice, and the number of these cells decrease significantly in Tbet- and Stat6-deficient mice. These results raise the possibility that IL-4 and TGF-β may simultaneously enhance the Th1 cell-mediated immune responses under certain cytokine conditions.

C57BL/6 and BALB/c mice were purchased from CLEA Japan. Stat6-deficient mice (25) on a C57BL/6 background were provided by Dr. S. Akira (Osaka University, Osaka, Japan). Tbet-deficient mice (6) were backcrossed to BALB/c mice (Charles River Laboratories) for eight generations. All mice were maintained under specific pathogen-free conditions and were used at 6–10 wk of age. All experiments using mice received approval from the Kazusa DNA Research Institute Administrative Panel for Animal Care. All animal care was conducted in accordance with the guidelines of Kazusa DNA Research Institute.

The anti–CD4-FITC mAb (RM4-5) was purchased from BD Biosciences, and the anti–CXCR3-allophycocyanin mAb (220803) was purchased from R&D Systems. The anti–CD38-PE mAb (90), anti–CD40-PE mAb (1C10), anti–CD96-PE mAb (3.3), anti–CD103-PE mAb (2E7), and anti-CD195 (CCR5)-Alexa Fluor 647 mAb (HM-CCR5) were purchased from BioLegend. For intracellular staining, an anti–IFN-γ–allophycocyanin mAb (XMG1.2; BD Biosciences), anti–IL-4–PE mAb (11B11; BD Bioscience), and anti–IL-9–PE mAb (RM9A4; BioLegend) were used.

The staining for the transcription factors was carried out with the Foxp3 Staining Buffer Set (eBioscience), according to the manufacturer’s instructions. The anti–Eomes-Alexa Fluor 647 mAb (Dan11mag) was purchased from both BD Biosciences and eBiosciences. The flow cytometric analysis was performed on a FACSCalibur instrument (BD Biosciences), and the results were analyzed with FlowJo software (Tree Star).

Naive CD4 (CD4^posCD62L^high) T cells were prepared using a CD4⁺CD62L⁺ T cell isolation Kit II (Miltenyi Biotec) and yielded a purity of CD4⁺CD62L^highCD44^low naive CD4 T cells >96%. Purified naive CD4 T cells (1.5 × 10⁶) were stimulated for 2 d with immobilized anti–TCR-β mAb (H57-597; 3 μg/ml) plus an anti-CD28 mAb (37.51; 1 μg/ml; BioLegend) in the presence of IL-2 (1 ng/ml) under the indicated culture conditions for 2 d; cells were transferred into new plates and cultured under the indicated cytokine conditions without anti-TCR/CD28 stimulation for an additional 3 d. The Th1 conditions were as follows: IL-12 (1 ng/ml) and an anti–IL-4 mAb (11B11; 1 μg/ml; BioLegend). The Th2 conditions were as follows: IL-4 (10 ng/ml) and an anti–IFN-γ mAb (1 mg/ml; BioLegend). The Th9 conditions were as follows: IL-4 (10 ng/ml), TGF-β (3 ng/ml), and an anti–IFN-γ mAb (1 μg/ml; BioLegend). The CD103⁺ Th1 conditions were as follows: IL-4 (3 ng/ml), IFN-γ (10 ng/ml), TGF-β (3 ng/ml). The differentiated Th cells were subjected to intracellular staining, quantitative RT-PCR, ELISA, immunoblotting, chromatin immunoprecipitation (ChIP), or (formaldehyde-assisted isolation of regulatory elements (FAIRE) analyses.

The pMX-IRES-hNGFR plasmid was generated, as described previously (26). Retrovirus vectors containing dominant-negative Eomes (Eomes-DN) cDNA (27) or active Stat6 cDNA (28) were used. An Eomes-DN cDNA construct was kindly provided by Dr. Steven L. Reiner (Abramson Family Cancer Research Institute, Philadelphia, PA) (27). The method used for the generation of retrovirus supernatant and the infection processes were described previously (29). Infected cells were detected by staining with an anti-human NGFR mAb (ME20.4-1.H4; BD Biosciences), and, where indicated, hNGFR⁺-infected cells were purified using a FACSAria cell sorter (BD Biosciences), yielding a purity > 98%.

Naive CD4 T cells were cultured under CD103⁺ Th1 conditions for 2 d, and the cells were subsequently infected with a retrovirus vector containing control (pLMP-IRES-hNGFR) or Eomes short hairpin RNA (shRNA) (pLMP-shEomes-IRES-hNGFR). An Eomes-containing microRNA-adapted retroviral vector, MSCV/LTRmiR30-PIG (LMP) (Open Biosystems), was used for the Eomes shRNA. Three days postinfection, hNGFR⁺ infected cells were purified and restimulated with an anti–TCR-β mAb for 4 h.

Total RNA was isolated using TRIzol reagent (Invitrogen). The cDNA was synthesized using a Superscript VILO cDNA synthesis Kit (Invitrogen). Quantitative RT-PCR was performed, as described previously, using an ABI Prism 7000 Sequence Detection System (Applied Biosystems). The primers and TaqMan probes used for the detection of Eomes, Il2, and Cd103 were purchased from Applied Biosystems. The expression was normalized to that of Hprt. The specific primers and Roche Universal Probes used in the experiments were as follows: Hprt forward, 5′-TCCTCCTCAGACCGCTTTT-3′, Hprt reverse, 5′-CCTGGTTCATCATCGCTAATC-3′, Hprt probe, #95; Tbet forward, 5′-TCAACCAGCACCAGACAGAG-3′, Tbet reverse, 5′-AAACATCCTGTAATGGCTTGTG-3′, Tbet probe, #19; Gata3 forward, 5′-TTATCAAGCCCAAGCGAAG-3′, Gata3 reverse, 5′-TGGTGGTGGTCTGACAGTTC-3′, Gata3 probe, #108; Pu.1 forward, 5′-GGAGAAGCTGATGGCTTGG-3′, Pu.1 reverse, 5′-CAGGCGAATCTTTTTCTTGC-3′, PU.1 probe, #94; Ifnγ forward, 5′-ATCTGGAGGAACTGGCAAAA-3′, Ifnγ reverse, 5′-TTCAAGACTTCAAAGAGTCTGAGGTA-3′, Ifnγ probe, #21; Il4 forward, 5′-CATCGGCATTTTGAACGAG-3′, Il4 reverse, 5′-CGAGCTCACTCTCTGTGGTG-3′, Il4 probe, #2; Il5 forward, 5′-GCCACTGCCATGGAGATT-3′, Il5 reverse, 5′-GGACAGGAAGCCTCATCG-3′, Il5 probe, #99; Il13 forward, 5′-CCTCTGACCCTTAAGGAGCTTAT-3′, Il13 reverse, 5′-CGTTGCACAGGGGAGTCT-3′, Il13 probe, #17; Cd38 forward, 5′-AAGATGTTCACCCTGGAGGA-3′, Cd38 reverse, 5′-CTCCAATGTGGGCAAGAGAC-3′, Cd38 probe, #3; Cd40 forward, 5′-AAGGAACGAGTCAGACTAATGTCA-3′, Cd40 reverse, 5′-AGAAACACCCCGAAAATGGT-3′, Cd40 probe, #105; Cd96 forward, 5′-CCACCTAGGTTCACCTTTTCA-3′, Cd96 reverse, 5′-GTGATGGTGGGTGAAGAGAAC-3′, Cd96 probe, #22; Ecadherin (Cdh1) forward, 5′-CCACCTAGGTTCACCTTTTCA-3′, Ecadherin (Cdh1) reverse, 5′-ACCACCGTTCTCCTCCGTA-3′, Ecadherin (Cdh1) probe, #18; Cxcr2 forward, 5′-CAGGACCAGGAATGGGAGTA-3′, Cxcr2 reverse, 5′-TCCCCTCCAAATATCCCCTA-3′, Cxcr2 probe, #32; Cx3cr1 forward, 5′-AAGTTCCCTTCCCATCTGCT-3′, Cx3cr1 reverse, 5′-CAAAATTCTCTAGATCCAGTTCAGG-3′, Cx3cr1 probe, #10; Ccr2 forward, 5′-ACCTGTAAATGCCATGCAAGT-3′, Ccr2 reverse, 5′-TGTCTTCCATTTCCTTTGATTTG-3′, Ccr2 probe, #27; and Ccr8 forward, 5′-AGAAGAAAGGCTCGCTCAGA-3′, Ccr8 reverse, 5′-GGCTCCATCGTGTAATCCAT-3′, Ccr8 probe, #4.

A ChIP assay was performed, as described previously (30), using anti-acetyl histone H3K9/14 anti-sera (Millipore) and anti-trimethyl histone H3K4 anti-sera (LP Bio) or anti-Stat6 anti-sera (Santa Cruz). The specific primers and Roche Universal probes used in the experiments were as follows: IFN-γ #1 forward, 5′-ATGATTCTTAAACTGGCCTTCAA-3′, IFN-γ #1 reverse, 5′-GATCATGAGTGCGAGGTCAC-3′, IFN-γ #1 probe, #84; IFN-γ #2 (promoter) forward, 5′-TCAACCAGCACCAGACAGAG-3′, IFN-γ #2 (promoter) reverse, 5′-CCCTTTTTGCCCTTGTAATG-3′, IFN-γ #2 (promoter) probe, #106; IFN-γ #3 (exon 3) forward, 5′-GAGGAACGCTGACTACAGATGA-3′, IFN-γ #3 (exon 3) reverse, 5′-TTATGTAGCGATCCCAGTTGC-3′, IFN-γ #3 (exon 3) probe, #20; Eomes #1 forward, 5′-CAGGGAGGTGATCCAGTAGG-3′, Eomes #1 reverse, 5′-TGAATGCTGGCTTCAGAATG-3′, probe, #55; Eomes #2 forward, 5′-GAGACCACAGGCCAAGATG-3′, Eomes #2 reverse, 5′-GCACTTGTTCCAAACAATACCA-3′, Eomes #2 probe, #9; Eomes #3 (promoter) forward, 5′-CCCTCCCCTAGATTTGCAC-3′, Eomes #3 (promoter) reverse, 5′-TGAAGGCGTCTTTACAAGCTC-3′, Eomes #3 (promoter) probe, #4; Eomes #4 (exon 1) forward, 5′-ACACCTTCGGGAGCACCT-3′, Eomes #4 (exon 1) reverse, 5′-CTCGGAGCTCAGGCTGTC-3′, Eomes #4 (exon 1) probe, #70; IL-9 promoter forward, 5′-ACCCGACTATTTGAAGAGCATC-3′, IL-9 promoter reverse, 5′-TGAGTCACTTGACAAAGGCTGT-3′, IL-9 promoter probe, #72; and IL-4 promoter forward, 5′-TTGGTCTGATTTCACAGGAAAA-3′, IL-4 promoter reverse, 5′-GGCCAATCAGCACCTCTCT-3′, IL-4 promoter probe, #2.

Cytoplasmic and nuclear extracts were prepared using the NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific). The Abs used for the immunoblot analysis were anti-Eomes anti-sera (H-109; Santa Cruz), anti-Gata3 mAb (HG3-31; Santa Cruz), anti–T-bet mAb (39D; Santa Cruz), anti-lamin A/C anti-sera (H-110; Santa Cruz), anti–E-cadherin anti-sera (Cell Signaling), and anti–α-tubulin mAb (DM1A; Lab Vision).

The levels of cytokine production were assessed by ELISA, as described previously (31).

FAIRE was performed, as described previously (32). In brief, 2 × 10⁶ cells were fixed with 1% paraformaldehyde at 37°C for 10 min. Then, cells were sedimented, washed, and lysed with 200 μl lysis buffer (2% Triton X-100, 1% SDS, 100 mM NaCl, 10 mM Tris-HCl [pH 8], 1 mM EDTA). The lysates were sonicated to reduce the DNA lengths to between 200 and 500 bp. DNA was purified by extraction with phenol/chloroform, following ethanol precipitation.

The small or large intestines were removed from normal or Tbet- or Stat6-deficient mice and separated from Peyer’s patches, and the intestines were cut longitudinally and into pieces 5 mm in length. The pieces were shaken for 40 min in magnesium-free, calcium-free HBSS (Life Technologies) supplemented with 1 mM DTT and 5% FCS. The cells were collected from the washes and passed over a discontinuous 40–70% Percoll gradient (Pharmacia Biotech) for 20 min at 900 × g, and the IEL were isolated from the Percoll-gradient interface.

The Student t test was used for statistical analyses.

As previously reported, the differentiation of Th2 cells is induced by IL-4, whereas the combination of IL-4 and TGF-β promotes Th9 cell generation in the absence of IFN-γ (2, 33). The supplementation of IFN-γ under Th9 conditions in culture inhibited the generation of Th9 cells (Fig. 1A, left panel). In addition, we found that IFN-γ–producing Th1 cells were markedly induced (Fig. 1A, left panel). The induction of Th1 cells by IFN-γ was modest under Th2 conditions in comparison with under Th9 conditions (Fig. 1A, right panel). Exposure to TGF-β in the absence of IL-4 showed a modest effect on IFN-γ–induced Th1 cell generation (Supplemental Fig. 1A). The augmentation of IFN-γ production in IFN-γ–treated Th9 cells, but not in IFN-γ–treated Th2 cells, was confirmed by ELISA (Fig. 1B, left panel). The production of IL-9 by Th9 cells was markedly inhibited by the IFN-γ supplementation (Fig. 1B, middle panel), whereas IL-4 production by Th2 cells was decreased about half by treatment with IFN-γ (Fig. 1B, right panel).

To characterize the cytokine-production profile, we compared the cytokine production in IFN-γ–treated Th9 cells with that in IL-12–induced Th1 cells. As shown in Fig. 1C, the production of IFN-γ and IL-2 from IFN-γ–treated Th9 cells was almost equivalent to that from IL-12–induced Th1 cells. IFN-γ–treated Th9 cells failed to produce Th2 cytokines, such as IL-4, IL-5, and IL-13 (Fig. 1C). The expression levels of cytokines, such as Tnfα, Il17a, Il17f, Il21, and Il22, in IFN-γ–treated Th9 cells were comparable to those in IL-12–induced Th1 cells (Supplemental Fig. 1B). To assess the changes in the chromatin status at the IFN-γ gene locus in IFN-γ–treated Th9 cells, we next performed a FAIRE assay (32). The FAIRE assay can be used for positive selection for the open regions of genomic DNA that are associated with regulatory activity, including those traditionally detected by the DNase I hypersensitivity assay. As expected, the supplementation of IFN-γ induced the open chromatin conformation at the IFN-γ promoter in CD4 T cells cultured under Th9 conditions but not Th2 conditions (Fig. 1D, left panel). The chromatin at the IL-9 promoter was closed by the treatment with IFN-γ (Fig. 1D, middle panel). The chromatin at the IL-4 promoter was also closed by IFN-γ administration in Th9 cells, whereas the status of IL-4 chromatin in Th2 cells was unaffected (Fig. 1D, right panel).

Next, we performed a DNA microarray analysis to identify the transcription factors that are important for IFN-γ production from IFN-γ–treated Th9 cells (Table I). As expected, high-level expression of T-bet, a Th1-related transcription factor, was induced by IFN-γ supplementation (Fig. 1E, left panel). In addition, Eomes, another T-box transcription factor, was highly expressed in IFN-γ–treated Th9 cells (Fig. 1E, middle panel). The expression of Eomes was not induced in IFN-γ–treated Th2 cells (Fig. 1E, middle panel). Another transcription factor, Pu.1, which is important for Th9 cell differentiation, was decreased by IFN-γ treatment (Fig. 1E, right panel). These results demonstrate that the cooperative stimulation by IL-4 and TGF-β (Th9 conditions) enhances IFN-γ–induced Th1 cell differentiation through induction of both Eomes and T-bet.

IFN-γ, IL-4, and TGF-β were added immediately or 2 d after TCR stimulation to determine the timing of the cytokine requirement in IL-4/TGF-β/IFN-γ–induced Th1 cell (IFN-γ–treated Th9 cells) development. IFN-γ was important during the initial activation phase of CD4 T cells, whereas TGF-β was required throughout the differentiation period (Fig. 1F, left and middle panels). IL-4 played a role during the late phase of differentiation (Fig. 1F, right panel). Hence, the cells were treated with IL-4 2 d after the initial TCR stimulation to examine the mechanism underlying the IL-4/TGF-β–induced augmentation of Th1 cell generation.

Next, we performed a DNA-array analysis and compared the mRNA expression levels of these cells with those of IL-12–induced Th1 cells (Table II). The expression profiles of integrins and chemokine receptors in IL-4/TGF-β/IFN-γ–induced Th1 cells were different from those of Th1 cells. As shown in Fig. 2A, the mRNA expression of Cd38, Cd40, Cd96, and Cd103 (ItgαE) was elevated in IL-4/TGF-β/IFN-γ–induced Th1 cells. The selective expression of Ecadherin mRNA was also detected (Fig. 2A). The protein expression levels of CD markers and E-cadherin were confirmed by flow cytometry (Fig. 2B) and immunoblotting (Fig. 2C), respectively. Because CD103 expression was stable, even after the removal of cytokines and subsequent restimulation with an anti–TCR-β mAb, we named the IL-4/TGF-β/IFN-γ–induced Th1 cells “CD103⁺ Th1 cells” (Fig. 2D). Although the preferential expression of CD38, CD40, and CD96 was also detected in CD103⁺ Th1 cells (Fig. 2B), the expression of these molecules was dependent on the cell-activation status (data not shown). The majority of Th9 cells, as well as a certain population of Th17 cells and iTreg, also expressed CD103, indicating the contribution of TGF-β to the regulation of CD103 expression (Supplemental Fig. 2A).

The expression of mRNA for chemokine receptors also showed a unique pattern. The expression of Cxcr2, Ccr2, and Ccr5 were low in CD103⁺ Th1 cells, whereas the levels of Cx3cr1 and Ccr8 were significantly higher (Fig. 2E). Low-level CCR5 protein expression was confirmed by flow cytometry (Supplemental Fig. 2B). Although the mRNA level of Cxcr3 in CD103⁺ Th1 cells was almost equivalent to that of Th1 cells (Fig. 2E), cell surface expression of CXCR3 was not detected in CD103⁺ Th1 cells (Supplemental Fig. 2B).

mRNA expression of Tbet, the master regulator of Th1 cell differentiation, was about half of the level in CD103⁺ Th1 cells in comparison with IL-12–induced Th1 cells (Fig. 3A, middle panel), whereas Eomes was predominantly expressed in CD103⁺ Th1 cells (Fig. 3A, left panel). The expression of Gata3 was lower in CD103⁺ Th1 cells compared with Th2 cells (Fig. 3A, right panel), although IL-4 was present in the CD103⁺ Th1 cell cultures. The protein expression of Eomes, T-bet, and Gata3 was confirmed by an immunoblotting analysis, and similar results were obtained as noted for the expression of mRNA (Fig. 3B). Although T-bet expression was lower in CD103⁺ Th1 cells on day 5 of culture, the level during the early-activation phase (day 1) was comparable to that in IL-12–induced Th1 cells (Fig. 3C, left panel). In sharp contrast, Eomes was detected in CD103⁺ Th1 cells during late differentiation cultures (Fig. 3C, right panel). To determine the role of Eomes in CD103⁺ Th1 cell development, we next introduced Eomes-DN into developing CD103⁺ Th1 cells. As shown in Fig. 3D, the generation of IFN-γ–producing cells was substantially reduced by the introduction of Eomes-DN. The decreased expression of Ifnγ was confirmed by quantitative RT-PCR analysis (Fig. 3E). Furthermore, the knockdown of Eomes by shRNA reduced the expression of Ifnγ in CD103⁺ Th1 cells (Fig. 3F). These data suggest that Eomes is induced during the late phase of CD103⁺ Th1 cell differentiation and is involved in the production of IFN-γ induced by TCR restimulation.

T-bet is a crucial transcription factor for Th1 cell differentiation (6) and is also expressed during CD103⁺ Th1 cell development (Fig. 3C). Therefore, we examined the role of T-bet in CD103⁺ Th1 cell differentiation. As indicated in Fig. 4A, Tbet deficiency resulted in a marked reduction in the generation of CD103⁺ Th1 and IL-12–induced Th1 cells. Stat4-deficient naive CD4 T cells differentiated normally into CD103⁺ Th1 cells, whereas the development of IL-12–induced Th1 cells was completely abolished (data not shown). The decreased expression of Ifnγ mRNA in Tbet-deficient CD103⁺ Th1 cells was confirmed by quantitative RT-PCR analysis (Fig. 4B). Of note, the level of IL-2 was unaltered in the Tbet-deficient CD103⁺ Th1 cells (Fig. 4B). The trimethylation level of histone H3K4 at the IFN-γ promoter was significantly reduced in the Tbet-deficient CD103⁺ Th1 cells, as well as in the IL-12–induced Th1 cells (Fig. 4C). Furthermore, T-bet was required for the formation of the open chromatin conformation at the IFN-γ gene locus in CD103⁺ Th1 cells (Fig. 4D). The expression of Eomes mRNA was also significantly decreased in the Tbet-deficient CD103⁺ Th1 cells, but the level was still higher in comparison with that in Th1 cells (Fig. 4E). The reduced expression of Eomes was confirmed at the protein level by an immunoblotting analysis (Fig. 4F).

We performed a FAIRE assay to obtain a better understanding of T-bet’s role in the expression of Eomes in CD103⁺ Th1 cells. As shown in Fig. 4G, the chromatin status at the Eomes gene locus was relatively closed in the Tbet-deficient CD103⁺ Th1 cells. This suggests that T-bet is involved in the formation of open chromatin at both the IFN-γ gene locus and the Eomes gene locus in CD103⁺ Th1 cells.

To assess the role of TGF-β in CD103⁺ Th1 cell development, we depleted TGF-β from the CD103⁺ Th1 cell cultures. As indicated in Fig. 5A, the removal of TGF-β resulted in the generation of IL-4–producing Th2 cells. Increased expression of Il4, Il5, and Il13 mRNA was also detected by quantitative RT-PCR analysis (Fig. 5B). We found that TGF-β was also involved in the expression of Eomes in CD103⁺ Th1 cells (Fig. 5C). The chromatin conformation at the Eomes gene locus was relatively closed in CD4 T cells cultured under TGF-β–depleted CD103⁺ Th1 conditions (Fig. 5D). In addition, TGF-β was found to be essential for the induction of CD103 expression (Fig. 5E). These data demonstrate that TGF-β is required for the induction of the CD103⁺ Th1 phenotype, as well as inhibition of Th2 cytokine expression.

Finally, we examined the role of Stat6 to assess the molecular mechanism underlying the IL-4–mediated augmentation of Th1 cell differentiation. The augmentation of CD103⁺ Th1 cell development by IL-4 was completely abolished in Stat6-deficient CD4 T cells, whereas IL-12–dependent Th1 cell differentiation was unaffected (Fig. 6A). The elimination of the IL-4–induced enhancement of Ifnγ expression was confirmed by quantitative RT-PCR analysis (Fig. 6B). The forced expression of active STAT6 (Stat6-VT) was sufficient to induce IFN-γ–producing Th1 cells cultured under IL-4–depleted CD103⁺ Th1 conditions (Fig. 6C). The open chromatin conformation at the IFN-γ gene locus was maintained in Stat6-deficient CD103⁺ Th1 cells (Supplemental Fig. 3A). In addition, H3K4 trimethylation at the IFN-γ gene locus showed only marginal decreases in Stat6-deficient CD103⁺ Th1 cells (Supplemental Fig. 3B). These results demonstrate that the IL-4/Stat6–signaling pathway is not crucial for chromatin remodeling at the IFN-γ gene locus in CD103⁺ Th1 cells.

The expression of Eomes was also substantially decreased in Stat6-deficient CD4 T cells cultured under CD103⁺ Th1 conditions, whereas the mRNA expression of Tbet was not significantly changed (Fig. 6D). Reduced expression of the Eomes protein in Stat6-deficient CD103⁺ Th1 cells was also detected by an immunoblotting analysis (Fig. 6E). Although the transcription of Eomes was decreased, the chromatin status was not changed, and high-level histone H3K4 trimethylation was still observed in Stat6-deficient CD103⁺ Th1 cells (Supplemental Fig. 3C, 3D). The forced expression of Stat6-VT enhanced Eomes expression (Fig. 6F), indicating that IL-4 augments Th1 cell generation via the STAT6-signaling pathway under CD103⁺ Th1 cell culture conditions.

To further evaluate the role of Stat6 in the expression of Eomes, we examined the direct binding of Stat6 to the Eomes gene locus by a ChIP assay. We checked several possible Stat-binding sites around the Eomes gene locus and found that Stat6 could bind to the exon 1 region in CD103⁺ Th1 cells; the binding of Stat6 at this region was not detected in Th2 cells (Fig. 6G). These results suggest that Stat6 is not essential for the formation of active chromatin at the Eomes gene locus but that it is required for the transcription of Eomes.

CD103 is expressed by >90% of IEL, and CD103 facilitates the retention of effector/memory lymphocytes in the gut epithelial layer via its interaction with E-cadherin (34–36). Therefore, we chose to use IEL to detect CD103⁺ Th1 cells in vivo. Fig. 7A shows that IFN-γ–producing CD103⁺ Th1 cells were observed in the small and large intestinal CD4⁺ IEL fraction but not in splenic CD4 T cells. These data clearly demonstrate the existence of CD103⁺ Th1 cells in vivo. The number of IFN-γ–producing CD103⁺ Th1 cells in small intestinal IEL decreased significantly in Tbet-deficient mice (Fig. 7B, left and middle panels). The total number of CD103⁺ CD4 T cells decreased as well (Fig. 7B, right panel). The generation of IFN-γ–producing CD103⁺ Th1 cells was also dependent on Stat6. As shown in Fig. 7C, IFN-γ–producing CD103⁺ Th1 cells (left and middle panels), as well as the total CD103⁺ CD4 T cells (right panel), decreased significantly in small intestinal IEL from Stat6-deficient mice. The reduced expression of Ifnγ mRNA in Stat6-deficient CD103⁺ CD4 T cells was confirmed by quantitative RT-PCR analysis (Fig. 7D). A decreased number of IFN-γ–producing CD103⁺ Th1 cells was also observed in IL-4–deficient mice (Supplemental Fig. 4). The high level of expression of Eomes mRNA was detected in small intestinal IEL, and this level decreased in Stat6-deficient mice (Fig. 7D). In contrast, the expression levels of mRNA for Tbet and Cd103 were unaffected (Figs. 7D, 8).

In this study, we identified a new pathway for Th1 cell development induced by the cooperative stimulation of IL-4, TGF-β, and IFN-γ. In addition, we demonstrated the IL-4/Stat6–dependent augmentation of Th1 cell development. Th1 cells induced by a combination of IL-4, TGF-β, and IFN-γ preferentially express CD103 (Fig. 2A, 2B); therefore, we termed these Th1 cells “CD103⁺ Th1 cells.” In the presence of IFN-γ, T-bet is induced and remodels chromatin at the IFN-γ gene locus and Eomes gene locus (Fig. 8, priming phase). TGF-β is also required for the formation of active chromatin at the Eomes gene locus (Fig. 8, priming phase). Then, Stat6, which is activated by IL-4, is recruited at the Eomes gene locus and induces transcription (Fig. 8, developing phase). TGF-β–mediated suppression of Th2 cell development is essential for this step (Fig. 8, developing phase). Because the IFN-γ gene locus has already been remodeled by T-bet, Eomes may permit the binding and induce the transcription of Ifnγ in CD103⁺ Th1 cells. In addition, TGF-β–mediated signal induces CD103 expression (Fig. 5E). Thus, a combination of IL-4, TGF-β, and IFN-γ induces the generation of CD103⁺ Th1 cells.

Yao et al. (37) reported that IL-4 induces Th1 cell differentiation though the modulation of IL-10 and IL-12 production from dendritic cells. In addition, IL-4, in combination with TGF-β, activates an alternative pathway for Th1 cell development in CD4 T cells that is independent of IL-12 (38). However, the molecular events underlying the activation of this pathway remain largely unknown. We found that activation of the IL-4/Stat6–signaling pathway in CD4 T cells under IFN-γ/TGF-β conditions leads to the expression of Eomes, as well as subsequent CD103⁺ Th1 cell development (Fig. 8). We confirmed that Eomes is involved in the production of IFN-γ from CD103⁺ Th1 cells (Fig. 3D–F). The Stat6-dependent transcriptional activation of the Eomes gene was observed only when both TGF-β and IFN-γ are present (Fig. 1E). Both IFN-γ–induced T-bet and TGF-β were required for the formation of open chromatin at the Eomes locus and for the subsequent Eomes expression (Figs. 4, 5). We found that Stat6 binds at exon 1 of the Eomes gene in CD103⁺ Th1 cells but not in Th2 cells (Fig. 6G). Therefore, IL-4 can augment the Th1 cell-mediated immune response by directly influencing the signaling pathway in CD4 T cells, as well as through modulation of dendritic cell functions.

Yagi et al. (39) reported high levels of Eomes and IFN-γ secretion in Gata3-deficient CD4 T cells cultured under Th2 conditions. They concluded that Gata3 blocks the Runx3–Eomes–IFN-γ pathway, presumably through interaction with Runx3. Indeed, the level of Runx3 in the CD103⁺ Th1 cells was higher than that of IL-12–induced Th1 cells, and the enforced expression of dominant-negative Runx3 inhibits Ifnγ expression in CD103⁺ Th1 cells (M. Yamashita, unpublished observations). We found that Gata3 was lower in CD103⁺ Th1 cells, although IL-4 was present in CD103⁺ Th1 cell cultures (Fig. 3A, 3B). Stat6 was normally activated by IL-4 under CD103⁺ Th1 conditions (M. Yamashita, unpublished observations). These observations suggest that the activation of Stat6 in the absence of Gata3 expression results in Eomes expression. Furthermore, we recently reported that Eomes inhibits Gata3 function in memory Th2 cells (40). Therefore, Eomes expressed in CD103⁺ Th1 cells may be involved in the suppression of Th2 cytokine production. Collectively, the Runx3–Eomes–IFN-γ pathway might be activated and involved in the establishment of the Th1 phenotype in CD103⁺ Th1 cells.

The cytokine-production profile of CD103⁺ Th1 cells was similar to that of IL-12–induced Th1 cells (Fig. 1C, Supplemental Fig. 1B), whereas the expression pattern of the adhesion molecules and chemokine receptors was unique (Fig. 2). These results suggest that the tissue distribution of CD103⁺ Th1 cells may be different from that of IL-12–induced Th1 cells. CD103, which associates with the β7 integrin to form a complex, was first detected as a marker of intraepithelial CD8 T cells in the gut (34, 35). αEβ7 integrin has been implicated in the retention of memory T lymphocytes within epithelial tissues, such as intestinal epithelial and airway epithelial tissues via binding to E-cadherin (41). Interestingly, CD103⁺ Th1 cells also express E-cadherin, a ligand for the αEβ7 integrin (Fig. 2A, 2C). Indeed, we found a CD103⁺ IFN-γ–producing CD4 T cell population in the intestinal IEL (Fig. 7A). IFN-γ production in these cells is dependent on both T-bet and Stat6 (Fig. 7B, 7C). CD103⁺ Th1 cells in small intestinal IEL expressed Eomes mRNA, and Stat6 was involved in its expression (Fig. 7D). Thus, these results suggest that CD103⁺ Th1 cells may play an important role in the immune responses in the epithelial mucosa.

Although CD103⁺ Th1 conditions (IL-4/IFN-γ/TGF-β) are similar to Th9 conditions (IL-4/TGF-β/anti–IFN-γ), IL-9 production was not detected in CD103⁺ Th1 cells (Fig. 1A, 1B). It was reported that IFN-γ negatively regulates IL-9 production (42). In addition, Th9 cells generated in vitro are easily converted into IFN-γ–producing cells after adoptive transfer into recipient mice, while maintaining their production of IL-9 (43). Moreover, Th9 cells mainly produced IL-17 in a colitis model induced by adoptive transfer (44), suggesting that Th9 cells can shift their cytokine profile toward Th1 or Th17 cells. We also found that IFN-γ evoked closed chromatin status at the promoter region of the IL-9 gene (Fig. 1D, middle panel). Moreover, the expression of PU.1, which is important for Th9 cell differentiation (45), was slightly reduced in IFN-γ–treated Th9 cells (CD103⁺ Th1 conditions) (Fig. 1E).

It is generally accepted that IL-4 induces Th2 cell differentiation and inhibits Th1 cell generation. However, our data provide molecular evidence for an ambivalent effect of IL-4 on the development of Th2 and Th1 cells. A combination of IL-4, TGF-β, and IFN-γ led to the differentiation of Th1 cells that expressed CD103 on their surface. Thus, a mixture of differentiation factors can have a biological effect that may be in opposition to those of the single components. The differentiation of CD103⁺ Th1 cells is a good example of Th cell development that depends on the integration of signals induced by different cytokines. IL-4 and IFN-γ are effector cytokines that are predominantly secreted from Th cells at the late phase of inflammation, and TGF-β is believed to be involved in chronic inflammation. This raises the possibility that under certain conditions, such as chronic inflammation, IL-4 may simultaneously enhance Th1 cell-mediated immune responses in cooperation with TGF-β.

We thank Yasuyo Tanaka, Saori Tsuda, Hikari Asou, Kaoru Sugaya, Miki Kato, and Hiroshi Sato for excellent technical assistance.

The authors have no financial conflicts of interest.