Runx1 and RORγt Cooperate to Upregulate IL-22 Expression in Th Cells through Its Distal Enhancer

The characterization of CD4⁺ Th cell subsets and the defining of their immunological functions have greatly expanded our understanding of the pathogenesis of infectious and inflammatory diseases. Naive Th cells differentiate upon activation into multiple distinct subsets, including Th1, Th2, Th17, and regulatory T (Treg) cells. Recently, Th22 cells have also been identified, so named for their predominant expression of IL-22 (1–3). IL-22 is a member of the IL-10 cytokine family and is produced by activated CD4⁺ Th cells, notably Th17 and Th22 cells, as well as by NK and innate lymphoid cells (4). Unlike most cytokines, IL-22 works exclusively on epithelial tissues, including the lung, liver, kidney, pancreas, gut, and skin (5–7). IL-22 protects these tissues from damage and enhances their regeneration (8). Exposure to IL-22 leads to the production of antimicrobial peptides, chemokines, and mucins (9–11), and neutralization of IL-22 results in the exacerbation of bacterial infections (12), suggesting an important role in both the maintenance of epithelial barrier integrity and host defense. However, dysregulated expression of IL-22 is associated with various autoimmune and allergic diseases such as rheumatoid arthritis, inflammatory bowel disease, psoriasis, and asthma (3, 6).

Expression of IL-22 in CD4⁺ T cells can be potently induced by cytokines such as IL-6 and IL-23, together with TCR signaling, in a STAT3-dependent manner (2, 13, 14). IL-6 alone can induce IL-22 expression, and TGF-β is actually inhibitory to its expression, whereas both IL-6 and TGF-β are necessary for the induction of IL-17 expression (2, 15). IL-23 alone is not sufficient to induce IL-22 expression from naive T cells because they do not express the IL-23R (16, 17). However, upon TCR activation in the presence of IL-6, IL-23R expression is rapidly induced in naive T cells, and therefore, IL-6 increases the responsiveness to IL-23 (18). STAT3-deficient CD4⁺ T cells fail to produce IL-22, whereas ectopic expression of a constitutively active STAT3 induces both IL-22 and IL-23R expression (13, 14). Besides STAT3, transcription of the Il22 gene is also positively regulated by other transcription factors such as the retinoic acid–related orphan receptor γt (RORγt) (19, 20), aryl hydrocarbon receptor (Ahr) (18, 21), and B cell–activating transcription factor (BATF) (22) because T cells deficient in any of these are impaired in their ability to produce IL-22. BATF directly binds to the Il22 promoter and induces its expression. IL-22 expression is promoted by ligands of the Ahr such as 6-formylindolo (3, 2-b) carbazole (FICZ) (23), and Ahr is implicated in the Il22 promoter function (19). Among these transcription factors, RORγt also serves as the master regulator of Th17 development and is both necessary and sufficient for Il17 expression (24).

Distal cis-regulatory elements, such as enhancers, silencers, and insulators, are required for proper lineage-specific gene expression and are often highly conserved across species. Multispecies sequence comparison and DNase I hypersensitivity site analysis have identified numerous elements in several cytokine loci (25). In Th2 cells, the expression of effector cytokine genes, including Il4, Il5, and Il13, is coordinately controlled by enhancers, a silencer, and a locus control region located in Rad50 introns (26–28). The mouse IFN-γ (Ifng) gene, which encodes the signature Th1 cytokine IFN-γ, is controlled by multiple cis-regulatory elements, including several enhancers and three CCCTC-binding factor (CTCF)–dependent insulators (29–32). In Th1 development, the distal cis-regulatory elements interact with the Ifng promoter by forming a chromatin loop, which therefore provides a platform capable of initiating Ifng transcription by bringing together transcription factors to a close spatial proximity (31). The mouse Il22 gene is located in the same Th1 cytokine locus as Ifng, whereas the human Th1 cytokine locus contains IFNG and two members (IL22 and IL26) of the IL-10 cytokine family (33). In some mouse strains (C57BL/6 and 129/Sv), the position occupied by IL26 in humans instead contains an inverted and nonexpressed duplication of Il22 (Iltifb) (34). Despite extensive characterization of the Ifng locus, little is known regarding the distal cis-regulatory elements that might control proper Il22 expression and Th22 differentiation.

The Runt-related transcription factor (Runx) gene family contains three members: Runx1, Runx2, and Runx3 (35). They bind to a specific DNA sequence through a highly conserved Runt domain. Runx proteins form heterodimers with the common partner, core-binding factor β (CBFβ), which protects them from ubiquitin-mediated degradation and induces conformational changes in their Runt domains that enhance DNA binding (36). The Runx/CBFβ complex interacts with other transcription factors, coactivators, or corepressors to regulate gene expression through binding to target promoters or enhancers. Among these Runx proteins, Runx1 and Runx3 have crucial functions in the establishment of lineage specification of T cells (37, 38). Runx1 binds to the promoter of Il2 and Ifng and induces the production of IL-2 and IFN-γ (39). Runx1 also has a significant role in Th17 differentiation because of its ability to induce RORγt expression and associate and act together with RORγt to induce Il17 expression (40). Runx3 acts cooperatively with the Th1-specific T box transcription factor (T-bet) to promote Ifng expression and silence Il4 in Th1 cells (41). The Runx1–RORγt interaction has a positive effect on Il17 expression, but the role of Runx1 in regulating Il22 expression remains unknown.

As a first approach to decipher the mechanism controlling Il22 expression in Th22 cells, we identified a novel enhancer element, termed conserved noncoding sequence (CNS)–32, in the mouse Il22 locus. Promoter-reporter assays that coupled this element to the Il22 promoter demonstrated substantially upregulated Il22 expression. CNS-32 contains binding motifs for Runx1 and RORγt, and we found that the recruitment of these transcription factors to CNS-32 synergistically induces IL-22 production in Th22 cells. Collectively, these results identify Runx1 as a novel positive regulator of Il22 expression and expand our understanding of the regulatory mechanism of Il22 expression with our discovery of the crucial Il22 enhancer that is controlled by Runx1 and RORγt.

C57BL/6 and BALB/c mice 6–12 wk of age were housed in specific pathogen–free conditions at the animal facility of Yamagata University and Fukushima Medical University. All animal experiments were performed in accordance with approved protocols from the institutional animal care and use committee of Yamagata University and Fukushima Medical University.

Total CD4⁺ T cells were isolated from mouse spleens by negative selection using anti-CD8α–biotin, anti-CD11b–biotin, anti-CD19–biotin, anti-CD25–biotin, anti-CD45R–biotin, anti-CD49b–biotin, and anti-TER119–biotin (BioLegend), followed by incubation with streptavidin-coated magnetic beads (BD Biosciences). Naive CD4⁺ T cells were further purified by flow cytometric sorting on a FACSAria II instrument for CD4⁺, CD25⁻, CD62L^high, and CD44^low surface marker expression. The culture medium used was IMDM supplemented with 10% FCS, 1 mM sodium pyruvate, nonessential amino acids, 50 μM 2-ME, 2 mM l-glutamine, and antibiotics (100 U/ml penicillin and 100 μg/ml streptomycin). For Th cell differentiation, naive CD4⁺ T cells were stimulated for 5 d with plate-bound anti-CD3 (1 μg/ml) and anti-CD28 (1 μg/ml) under several Th cell conditions as follows: Th1 cells: 10 ng/ml IL-12 and 10 μg/ml anti–IL-4 (11B11) (BioLegend); Th2 cells: 10 ng/ml IL-4 and 10 μg/ml anti–IFN-γ (XMG1.2) (BioLegend); Th17 cells: 40 ng/ml IL-6, 2 ng/ml TGF-β, 20 ng/ml IL-23, 10 μg/ml anti–IFN-γ, and 10 μg/ml anti–IL-4; and Th22 cells: 40 ng/ml IL-6, 20 ng/ml TNF-α, 20 ng/ml IL-1β, 20 ng/ml IL-23, 10 μg/ml anti–IFN-γ, 10 μg/ml anti–IL-4, 5 μg/ml anti–TGF-β1/2/3, and 300 nM FICZ (R&D Systems). NK cells were purified by flow cytometric sorting as having the surface phenotype CD3⁻DX5⁺NKp46⁺ among cells from BALB/c mice and were expanded for 5 d under 10 ng/ml IL-2 and 10 ng/ml IL-18 (MBL International). Mouse EL4 and human Jurkat cells were cultured in RPMI 1640 supplemented with 10% FCS and antibiotics. HEK293T and Plat-E cells were cultured in DMEM supplemented with 10% FCS and antibiotics.

For cytokine production, cells were stimulated for 5 h with 50 ng/ml PMA and 500 ng/ml ionomycin (Calbiochem) in the presence or absence of GolgiPlug (BD Biosciences). Cells were stained for surface Ags and then treated with Cytofix/Cytoperm (BD Biosciences) for intracellular staining. The following Abs were used: CD271-allophycocyanin, IL-17A–PE, and IL-22–PE (BioLegend). Flow cytometric analysis was performed on a FACSCanto II instrument (BD Biosciences). IL-22 in culture supernatants was measured with the ELISA kit (eBioscience).

Total RNA was extracted with Isogen (Nippon Gene), and cDNA was generated with the PrimeScript RT Master Mix (Takara). Real-time quantitative RT-PCR (qRT-PCR) was performed with the KAPA SYBR FAST qPCR kit (KAPA Biosystems) and the following primers: Ahr, 5′-AGCCGGTGCAGAAAACAGTAA-3′ and 5′-AGGCGGTCTAACTCTGTGTTC-3′; Batf, 5′-CTGGCAAACAGGACTCATCTG-3′ and 5′-GGGTGTCGGCTTTCTGTGTC-3′; Cbfb, 5′-GGAGTTTGATGAGGAGCGAG-3′ and 5′-GGTCTTGCTGTCTTCTTGCC-3′; Gata3, 5′-CTCGGCCATTCGTACATGGAA-3′ and 5′-GGATACCTCTGCACCGTAGC-3′; hypoxanthine phosphoribosyltransferase (Hprt), 5′-TCAGTCAACGGGGGACATAAA-3′ and 5′-GGGGCTGTACTGCTTAACCAG-3′; Ifng, 5′-ATGAACGCTACACACTGCATC-3′ and 5′-CCATCCTTTTGCCAGTTCCTC-3′; Il17a, 5′-TTTAACTCCCTTGGCGCAAAA-3′ and 5′-CTTTCCCTCCGCATTGACAC-3′; Il22, 5′-GACCAAACTCAGCAATCAGCTC-3′ and 5′-TACAGACGCAAGCATTTCTCA G-3′; Il23r, 5′-TTCAGATGGGCATGAATGTTTCT-3′ and 5′-CCAAATCCGAGCTGTTGTTCTAT-3′; Il4, 5′-TTTGGCACATCCATCTCCG-3′ and 5′-AGATCATCGGCATTTTGAACG-3′; Itga2, 5′-TACAGCAGCTTACGAACCCAC-3′ and 5′-CTGGTGAGGGTCAATCCCA-3′; c-Maf, 5′-AAATACGAGAAGCTGGTGAGCAA-3′ and 5′-CGGGAGAGGAAGGGTTGTC-3′; Runx1, 5′-GCAGGCAACGATGAAAACTACT-3′ and 5′-GCAACTTGTGGCGGATTGTA-3′; Runx3, 5′-CAGGTTCAACGACCTTCGATT-3′ and 5′-GTGGTAGGTAGCCACTTGGG-3′; Rora, 5′-GTGGAGACAAATCGTCAGGAAT-3′ and 5′-TGGTCCGATCAATCAAACAGTTC-3′; Rorc, 5′-TCCTGCCACCTTGAGTATAGTC-3′ and 5′-GTAAGTTGGCCGTCAGTGCTA-3′; Tbx21, 5′-AGCAAGGACGGCGAATGTT-3′ and 5′-GGGTGGACATATAAGCGGTTC-3′. Target gene expression was calculated using the comparative method for relative quantification after normalization to Hprt gene expression.

cDNA for Runx1, Runx3, Cbfb, Est1, Batf, Fosl2, and Rorc was cloned into the pcDNA3 vector (Thermo Fisher Scientific). The mouse Il22 promoter (−1427 to +21) or a minimal irrelevant promoter (minP) (5′-AGAGGGTATATAATGGAAGCTCGACTTCCA G-3′) was cloned upstream of the luciferase gene in the pGL3-Basic vector (Promega), resulting in pGL3–IL22 promoter or pGL3–minP. The Ifng promoter-reporter construct was described previously (30). The CNS elements were PCR amplified and cloned either upstream of the promoter or downstream of the luciferase gene in the pGL3-Basic vector. The primers used were as follows: CNS-34, 5′-GTCCAAGATTCTGCTGTCC-3′ and 5′-GCTTTCATCCTGGCTGAGTC-3′; CNS-32, 5′-CCAGAGCACTCTGACCTCCC-3′ and 5′-TCCCGAGCATACATTCCATC-3′; CNS-25, 5′-TGCCTTTCCACAGGCCATGC-3′ and 5′-ATTGCTCCTCCTACTTCTAC-3′; CNS-18, 5′-AGGGTCTAAACTAGGAGTGG-3′ and 5′-TGTACATGACACCAGGTGCC-3′; Il22 promoter, 5′-GAGTGCTTTTAGATTCCACC-3′ and 5′-TCGCACAAGTGTCAACAGTTG-3′. Deletion constructs of CNS were generated by restriction enzyme digestion, and mutations in CNS were introduced by site-directed mutagenesis. Luciferase assays were performed in EL4 or Jurkat cells by electroporation using a Gene Pulser II instrument (Bio-Rad). Transfected cells were incubated overnight and left untreated or treated for 5 h with PMA and ionomycin. Luciferase activities were measured with the Dual-Luciferase reporter kit (Promega), normalized with a Renilla control, and shown as fold induction relative to the empty vector control.

MIG and MIN are bicistronic retroviral vectors containing GFP and human nerve growth factor receptor (hNGFR) lacking the cytoplasmic domain, respectively, under the control of an internal ribosome entry site (IRES). Runx1 cDNA was cloned into MIG (MIG-Runx1). Rorc cDNA was amplified by PCR with a 5′ Flag-tagged primer and was cloned into MIG (MIG-RORγt) or MIN (MIN-RORγt). For short hairpin RNA (shRNA) knockdown, dsDNA sequence targeting the coding region of Cbfb (5′-GCTCGAAGAAGAACTCGAGAA-3′) (42) or luciferase (5′-ACTTACGCTGAGTACTTCGAA-3′) was cloned into the MSCV-LTRmiR30-PIG (LMP) retroviral vector as described previously (31). The retrovirus-based reporter hNGFR-pA-GFP-RV, in which hNGFR marks viral infection and GFP is used to report promoter activity, was generated as described previously (43). Retroviral packaging was performed with Plat-E cells as described previously (44). Virus-containing supernatants were collected and supplemented with polybrene (8 μg/ml) before spinfection. Knockdown efficiency of CBFβ protein expression was assessed by immunoblotting with anti-CBFβ and anti-GAPDH (Santa Cruz Biotechnology).

DNA pull-down assays were performed following protocols as described previously, with minor modifications (45). Briefly, biotinylated DNA probes were conjugated to M-280 streptavidin beads (Thermo Fisher Scientific) and then blocked by 0.5% BSA in binding buffer (10 mM Tris-HCl, pH 8.0, 0.1 mM EDTA, 1 mM DTT, 100 mM NaCl, 10% glycerol). Nuclear extracts were prepared from HEK293T cells overexpressing Runx1 or RORγt with lysis buffer (10 mM HEPES, pH 7.9, 100 mM KCl, 5 mM MgCl₂, 10% glycerol, 0.1% Nonidet P-40, and 1 mM DTT) containing protease inhibitors and incubated with the DNA-conjugated beads for 2 h at 4°C in binding buffer containing 20 μg of poly(dI:dC). The beads were washed four times with the binding buffer, and bound proteins were detected by immunoblotting with anti-Runx1 (Santa Cruz Biotechnology) and anti-RORγt (eBioscience). Probe sequences are described in Fig. 4A.

Chromatin immunoprecipitation (ChIP) assays were performed as described previously (31). Chromatin prepared from T cells was immunoprecipitated using anti-Runx1 (Santa Cruz Biotechnology), anti–acetyl-histone H4 (Millipore), anti-Flag (Sigma-Aldrich), or normal rabbit IgG (Santa Cruz Biotechnology). DNA contents were measured by real-time quantitative PCR (qPCR) with the KAPA SYBR FAST qPCR kit and the following primers: CNS-34, 5′-GGAATAGCATGCAATTTGAGCAGA-3′ and 5′-GGCCTAGATTTAACTCACTGGACA-3′; CNS-32, 5′-AAAAGGGGCTCATTTTAGGG-3′ and 5′-GCAATCGCTGAGTCACAACA-3′; CNS-25, 5′-TGCCTTTCCACAGGCCATGC-3′ and 5′-AGCCCGTCTGTGCTAACTTC-3′; Il22 promoter, 5′-TTTCTGGGATTTGTGTGCAA-3′ and 5′-TTTATAGCCATCGGGACACC-3′; Il2 promoter, 5′-CACAACAGGCTGCTTACAGG-3′ and 5′-GTCGGGTTAGCCCACACTTA-3′; Il17a CNS2, 5′-CAGATGCATGCAGAACTGACT-3′ and 5′-AAGGCTCTGGAGAGCAGACA-5′; Actb promoter, 5′-CCCGCGTGTCCCTCAA-3′ and 5′-TCACGTGGATATCGAGCACTTAA-3′; Gfap intron 1, 5′-CTGCCGCCTTAGCTCATTAC-3′ and 5′-AACCAAGGGTCACACTCCAG-3′. An unrelated genomic region of the Gfap gene was used as a negative control. As a loading control, qPCR was performed directly on input DNA purified from chromatin before immunoprecipitation, and sample values were presented as a percentage of input.

Methylation analysis was performed as described previously (31). Samples were prepared by performing nested PCR to amplify the Il22 promoter and its enhancer. The primers used were as follows: Il22 promoter outside, 5′-GAATATTTTGTGGTAATGGGAGTTG-3′ and 5′-CCTAACCACCATAAAAAACCACAAA-3′; Il22 promoter nested, 5′-GGTTTATTTTAAAGTATAGAATATAGG-3′ and 5′-AAACCTTCCCACCTTAAAAAATAAA-3′; CNS-32 outside, 5′-AAGTTTATTTAGTTGTGAGTTTGTG-3′ and 5′-TAACTAACAACACACATAAAAAACC-3′; and CNS-32 nested, 5′-TGTTTGTAAAGGAGTTTTATATATGG-3′ and 5′-ACAAACTTACCAATCTTTCATCAAC-3′. The PCR product was cloned using the TOPO TA Cloning kit (Thermo Fisher Scientific). Plasmid DNA from individual clones was sequenced.

The chromosome conformation capture (3C) assay was performed as described previously (31), using BglII and BamHI for enzymatic digestion of chromatin prepared from Th1, Th22, and MEF cells. The 3C samples were analyzed using PrimeTime real-time qPCR (Integrated DNA Technology). For a control template, equimolar amounts of four BAC clones spanning the mouse Il22 locus (B6Ng01-263N13, B6Ng01-295L21, B6Ng01-161P12, and B6Ng01-333J24; obtained from RIKEN Bioresource Research Center) and a BAC spanning the mouse Actb locus (B6Ng01-258D04) were mixed and then digested and ligated. Ligation frequencies between the analyzed pairs were normalized to those detected between two restriction fragments in the Actb locus. The primers and PrimeTime probes used for real-time PCR were as follows: 3C-22CNS-32 (anchor), 5′-GTGGATGAGTCACTGGCAGA-3′; 3C-22CNS-18, 5′-GCACCCACAGATAGAACCCA-3′; 3C-22-7, 5′-GCCTGGTCAGTCATCTCCTG-3′; 3C-22-3, CCCAGGTCCCAGACAGACTA-3′; 3C-22 promoter, 5′-AGACTGACACGCAAATGCCT-3′; 3C-22+3, 5′-CTCAGTGGATATCGCCCCCT-3′; 3C-Actb-1, 5′-TGGCATTGTTACCAACTGGGA-3′; 3C-Actb-2, 5′-GTCTAAGTGGAGCCCCTGTC-3′; 22CNS-32–probe, 5′-FAM-CCATGTAGCAAGCAGCTACCCTTAAGGTG-IBFQ-3′; Actb-probe, 5′-FAM-AGACATGCAAGGAGTGCAAGAACACAGC-IBFQ-3′.

Statistical significance was calculated by the Mann–Whitney U test. Error bars represent the SD, as indicated. The p values <0.05 are considered significant.

Multispecies sequence comparison reveals evolutionally CNSs that often contain functional regulatory elements. Using the VISTA alignment program, we compared DNA sequences flanking the mouse Il22 gene with the corresponding human genomic sequence. Similar to those reported in the Il4 and Ifng loci (26, 29), we identified four CNSs upstream of Il22 with over 75% sequence identity between mouse and human genomes (Fig. 1A). These CNSs were located 34 kb (CNS-34), 32 kb (CNS-32), 25 kb (CNS-25), and 18 kb (CNS-18) upstream of the Il22 promoter. To determine whether any of the CNSs might have transcriptional regulatory activity, DNA fragments containing each CNS were cloned upstream of the Il22 promoter, and the luciferase reporter constructs were transfected into EL4 cells (Fig. 1B). Because C57BL/6 mice have the Iltifb gene region, which represents an inverted and nonexpressed duplication of Il22 and its surrounding 5′ and 3′ flanking sequences (Supplemental Fig. 1A), BALB/c mice were used for cloning the CNSs (34). The inverted duplication of the sequences, however, did not affect IL-22 expression at both protein and mRNA transcript levels in Th22 cells from C57BL/6 mice (Supplemental Fig. 1B, 1C). When CNS-32 was tested, marked enhancer activity was observed in response to PMA and ionomycin, whereas the other CNSs did not confer any further increase in reporter activity (Fig. 1B).

To identify an essential region in CNS-32 for its enhancer activity, we generated the truncated versions that were cloned in the Il22 promoter–reporter construct and applied these constructs in reporter assays. All constructs containing the fragment B (hereafter, CNS-32B) were capable of significantly enhancing reporter activity. In contrast, other constructs containing fragment A or C alone did not show any activity (Fig. 1C). When CNS-32B was cloned in the reverse orientation or downstream of the luciferase gene in the reporter construct, these constructs also showed a similar enhancement of reporter activity in EL4 cells (Fig. 1D). These results indicate that CNS-32B is essential for Il22 enhancer activity that functions in an orientation- and position-independent manner. Intriguingly, CNS-32B also significantly enhanced Ifng promoter activity in EL4 cells. Such reporter behaviors were recapitulated in human Jurkat cells (Fig. 1D).

To determine possible transcription factors required for Il22 enhancer activity, we performed sequence analysis of the identified CNS-32B region using the Genomatix program and found putative binding motifs for NF-κB, RORγt, Gfi1, Ets, Runx, and AP-1 (marked as AP-1a and AP-1b) transcription factors (Fig. 2A). Each of these motifs was mutated individually to assess their contribution to the Il22 enhancer function. We cloned the mutated CNS-32B fragments upstream of a minP in the reporter construct and then tested their ability to increase reporter activity in EL4 cells. Most strikingly, mutation of the AP-1a site largely abolished enhancer activity (Fig. 2B). In addition, Il22 enhancer activity was significantly reduced when the RORγt, Ets, Runx, and AP-1b sites were mutated, whereas mutation of the NF-κB and Gfi1 sites did not affect enhancer activity.

To determine if any of the transcription factors can confer enhancer activity, reporter assays were performed using EL4 cells cotransfected with the reporter construct along with empty vector or expression vectors encoding the candidate transcription factors (RORγt, Ets-1, Runx1, Runx3, BATF, and Fosl2). Ets-1 (Ets family member) and BATF and Fosl2 (AP-1 family member) were used because they are critical in Th17 cell differentiation and function (46, 47). Among the tested transcription factors, RORγt, Runx1, and Runx3 significantly enhanced reporter activity in response to PMA and ionomycin (Fig. 3A). Unexpectedly, Ets-1, BATF, and Fosl2 showed no significant enhancement of reporter activity. Runx1 was broadly expressed among all Th cell subsets, whereas Runx3 was only upregulated in naive CD4⁺ T cells cultured under Th1 conditions (Supplemental Fig. 2A). As previously reported, Runx3 may function as a key transcription factor for Th1 cells (41); therefore, we decided to focus on Runx1. Generally, the transcriptional activity of all Runx proteins is dependent on the formation of heterodimers with CBFβ (36). Thus, overexpression of Runx1 together with CBFβ strongly enhanced reporter activity, although CBFβ, when transfected alone, had no reporter activity enhancement (Fig. 3B). Furthermore, we observed the functional synergy between Runx1 and RORγt in the reporter assays. Overexpression of Runx1 together with RORγt synergistically conferred a strong luciferase signal, but this signal was much lower when two binding sites were mutated (Fig. 3C). However, the synergistic effect on the enhancement of reporter activity was not seen in other combinations of the selected transcription factors (Supplemental Fig. 3).

We finally examined whether the enhancer activity of CNS-32B depends on both Runx1 and RORγt in primary Th22 and NK cells. Similar to a previous report (43), we used a reverse-stranded retroviral reporter (hNGFR-pA-GFP-RV) to obtain high transfection efficiency. In this assay, enhancer activity will be reflected by the intensity of the GFP signal within hNGFR⁺ cells. Naive CD4⁺ T cells activated under Th22 conditions were transduced with various GFP reporter retroviruses. The wild-type (WT) CNS-32B–containing retroviral reporter showed strong GFP expression, whereas its expression was diminished when the binding sites for Runx1 and/or RORγt were mutated (Fig. 3D). This enhancer activity was also detected in NK cells where Il22 could be potentially expressed (Fig. 3E). Together, these results suggest that CNS-32 specifically functions as a Runx1- and RORγt-responsive enhancer for Il22 expression in IL-22–producing cells, including Th22 and NK cells.

To examine the physical binding of Runx1 and RORγt to CNS-32 in vitro, DNA pull-down assays were performed with DNA probes, including their target sites (Fig. 4A). We detected Runx1 binding to the Rx-WT probe but not to its mutant (Rx-mut) in which the Runx site was disrupted (Fig. 4B). Additionally, the pull-down efficiency was significantly reduced by consensus Runx probe (IL2Rx-WT) from the human IL2 promoter (39) but not by its mutant (IL2Rx-mut), indicating the specificity of Runx1 binding. RORγt binding to CNS-32 was also detected by the same assays. RORγt protein was effectively precipitated with biotinylated Rγt-WT probe but not its mutant (Rγt-mut) in which the RORγt site was disrupted (Fig. 4C). RORγt binding to CNS-32 was completely inhibited in the presence of consensus RORγt probe (IL17Rγt-WT) from the mouse Il17 CNS2 (40) but was not affected by its mutant (IL17Rγt-mut).

We further performed ChIP assays in differentiated Th cells to confirm the physical binding of Runx1 and RORγt to the predicted binding sites in CNS-32 in vivo. First, we examined the transcript expression of Th cell–associated cytokines and transcription factors using real-time qRT-PCR (Supplemental Fig. 2A). Th1 cells expressed the highest amounts of Ifng and Tbx21 transcripts; Th2 cells had the highest abundance of Il4 and Gata3; Th17 cells produced the greatest abundance of Il17a, Rora, and Rorc; and Th22 cells produced the highest amounts of Il22, demonstrating that these cells were successfully differentiated. We observed a significant recruitment of Runx1 to CNS-32 in Th22 cells (Fig. 4F). In contrast, the recruitment of Runx1 to CNS-32 in Th1 and Th17 cells was much lower than that in Th22 cells (Fig. 4D–F). We confirmed that Runx1 binds to the Il2 promoter as a known target (39) but not to the Gfap intron 1 as a negative control. Importantly, Runx1 binding to CNS-32 was specific because no significant Runx1 binding was observed at the Il22 promoter and other CNSs or in a control ChIP with Ig G. Because anti-RORγt Ab suitable to ChIP was not available, Flag-tagged RORγt was retrovirally overexpressed in naive CD4⁺ T cells cultured under Th22 conditions. ChIP assays using anti-Flag revealed a similar enrichment pattern to that observed for Runx1 in Th22 cells (Fig. 4G). No significant RORγt binding was observed at the Il22 promoter and the other CNSs or in a control IgG. Consistent with a previous report (40), we confirmed that RORγt binds to the Il17a CNS2 region as a known target. Collectively, these results suggest the intriguing possibility that Runx1 and RORγt may collaboratively regulate IL-22 expression through their binding to the Il22 enhancer in Th22 cells.

Epigenetic changes in the levels of histone acetylation and DNA methylation are well associated with chromatin remodeling and transcriptional activity. To determine whether these changes occur during Th22 differentiation, we cultured naive CD4⁺ T cells under Th1, Th17, or Th22 conditions and analyzed histone H4 acetylation (H4Ac), a typical feature of open chromatin, in the Il22 locus using ChIP assays. In naive CD4⁺ T cells, where Il22 was not expressed (Supplemental Fig. 2B), merely background levels of H4Ac were observed at the tested CNSs and promoter in the Il22 locus (Fig. 5A). Following Th22 differentiation, Il22 expression was greatly induced (Supplemental Fig. 2A, 2B), and increased levels of H4Ac were detected at most of the CNSs and promoter (Fig. 5D). However, compared with Th22 cells, H4Ac at these regions in the Il22 locus was significantly reduced in Th1 and Th17 cells (Fig. 5B, 5C). Consistent with β-actin expression in all Th cells, histone H4 molecules in the Actb promoter were highly acetylated (Fig. 5). Conversely, the first intron of Gfap, chosen as a negative control, exhibited no association with H4Ac in all cell samples. Importantly, higher levels of H4Ac were observed at CNS-32 identified as the Il22 enhancer, where Runx1 and RORγt were most highly enriched in differentiated Th22 cells (Fig. 4F, 4G).

We next analyzed the DNA methylation status of the CNS-32 and Il22 promoter by bisulfite sequencing. In naive CD4⁺ T cells, most CpG sites of CNS-32 were demethylated, whereas the promoter exhibited partial CpG demethylation (Fig. 5E). The pattern of CpG methylation in these regions in Th1 and Th17 cells was similar to that observed in naive CD4⁺ T cells (Fig. 5E). When naive CD4⁺ T cells were cultured under Th22 conditions, the two regions were highly demethylated, correlating with increased Il22 expression (Fig. 5E, Supplemental Fig. 2A). In contrast, the CpG sites of both regions were mostly methylated in MEF cells and hepatocytes (cell types that cannot express Il22 [Fig. 5E, Supplemental Fig. 2B]). These epigenetic analyses, together with gene expression data, suggest that the association of Runx1 and RORγt with CNS-32 might be necessary for the seeding and spreading of chromatin remodeling across the entire locus, providing accessibility of the Il22 promoter and other regulatory elements.

We next performed a 3C assay to investigate whether the Il22 locus is accompanied by any change in chromatin conformation during the induced expression of IL-22. We examined a total of six BglII and BamHI fragments, within which we designed reverse primers to screen for chromatin loops, including sites corresponding to CNS-32, CNS-18, the promoter region, and other intergenic sequences. When Th1 and MEF cells were subjected to the 3C assay with CNS-32 as an anchor, we did not observe any formation of chromatin loops across the Il22 locus, suggesting that the Il22 promoter is excluded from any regulatory elements in these cells (Fig. 6A). However, when naive CD4⁺ T cells were cultured under Th22 conditions to induce Il22 expression, a strong association of CNS-32 with the Il22 promoter was observed (Fig. 6B). These results reveal that IL-22 expression is accompanied by dynamic changes of chromatin conformation in the Il22 locus, bringing the CNS-32 enhancer in close proximity to the Il22 promoter.

To directly investigate whether Runx1 and RORγt promote IL-22 production in primary CD4⁺ T cells, we used retroviral transduction to deliver Runx1 or RORγt to Th22 cells. We observed that the percentage of cells producing IL-22 is significantly higher in cells expressing either Runx1 or RORγt alone than in control cells expressing GFP alone (Fig. 7A). Similar to the published report that Runx1 and RORγt are linked to IL-17 production in Th17 cells (40), as expected, ectopic overexpression of Runx1 or RORγt resulted in a marked increase in the percentage of Th22 cells producing IL-17 (Fig. 7A). We confirmed these observations using ELISA for IL-22 in the supernatants from Th22 cell cultures (Fig. 7B) and real-time qRT-PCR analysis (Fig. 7C). Moreover, the augmentation of Il22 expression by ectopic overexpression of Runx1 or RORγt was observed in Th22 cells but not in Th1 or Th17 cells (Fig. 7D). In addition, ChIP assays showed that the recruitment of Runx1 and RORγt to CNS-32 in Th22 cells was significantly higher than that in Th1 or Th17 cells (Fig. 7E). Thus, IL-22 expression levels were well correlated with the recruitment levels of Runx1 and RORγt to the Il22 enhancer.

To further understand the role of Runx1 and RORγt in directing Il22 expression, we sorted GFP⁺ cells from the retrovirus-transduced Th22 cells and assessed their gene expression profiles with real-time qRT-PCR analysis. Ectopic Runx1 expression greatly enhanced Il23r, Rora (encoding RORα), and Rorc (encoding RORγt) expression, whereas Runx3, Cbfb, and Ahr expression were not affected (Fig. 7C). In contrast, ectopic RORγt expression enhanced Il23r expression alone without affecting other genes’ expression. These results suggest that Runx1 and RORγt promote IL-22 production in Th22 cells by indirectly amplifying IL-23 signaling via the induction of Il23r expression.

Runx1 has been reported to promote IL-17 production by both inducing RORγt expression and binding to the RORγt protein (40). To examine whether Runx1 functions in cooperation with RORγt in IL-22 production, naive CD4⁺ T cells were cotransduced with two retroviruses expressing GFP alone (MIG), GFP-Runx1 (MIG-Runx1) and hNGFR alone (MIN), or hNGFR-RORγt (MIN-RORγt). We cultured the transduced cells under Th22 conditions and measured IL-22 production of gated GFP⁺NGFR⁺ cells by intracellular staining. Either Runx1 or RORγt alone, but not control retroviruses expressing only the marker genes, was sufficient to promote the generation of cells producing IL-22 (Fig. 8A). Remarkably, coexpression of Runx1 and RORγt led to greatly enhanced IL-22 production (Fig. 8A). We confirmed these findings using ELISA for IL-22 in the supernatants from the sorted GFP⁺NGFR⁺ cells and real-time qRT-PCR analysis, suggesting that Runx1 and RORγt synergistically promote IL-22 production in primary Th22 cells (Fig. 8B, 8C).

To further evaluate whether Runx1 is involved in promoting IL-22 expression, we knocked down CBFβ expression in Th22 cells using shRNA introduced via retroviral transduction. The ablation of CBFβ results in the complete loss of Runx function because transcriptional activity of all Runx proteins is dependent on CBFβ (36). A previously characterized CBFβ shRNA construct (42) produced marked reduction of CBFβ protein and mRNA levels (Fig. 9A, 9C). The reduction was directly paralleled by a reduction in the percentage of cells producing IL-22, Il22 mRNA abundance, and the amounts secreted into culture supernatants (Fig. 9B–D). We also confirmed decreased Il17a mRNA expression using CBFβ shRNA knockdown, similar to observations made in Th17 cells (40). Importantly, mRNA levels of Il23r, Runx1, Runx3, Rora, and Rorc remained unchanged, indicating the specificity of the Runx/CBFβ complex–dependent effects on Il22 expression. However, the recruitment of RORγt to the Il22 locus was not affected by CBFβ shRNA knockdown in Th22 cells (Fig. 9E), which suggests that the chromatin accessibility of RORγt is independent of the presence of the Runx/CBFβ complex. Collectively, these results support an important role of the Runx/CBFβ complex in the positive regulation of Il22 expression.

In this study, we have extended the functional map of the Il22 locus through delineation of interactions between distal cis-regulatory elements and key transcription factors that direct Il22 expression. We identified a highly conserved upstream element designated as CNS-32. This element resides in a region of histone hyperacetylation and DNA demethylation in Th22 cells, suggesting it may play an important role in regulating Il22 expression. CNS-32 contains the Runx1 and RORγt binding motifs. By using reporter assays, we showed that CNS-32 robustly enhanced reporter activity depending on these binding motifs. We found that Runx1 and RORγt bind to CNS-32 during the course of Th22 differentiation. Importantly, retroviral overexpression of Runx1 and RORγt in Th22 cells synergistically promoted IL-22 production. Consequently, our study defines Runx1 as a novel positive regulator of Il22 expression, and we thus propose that lineage-specific IL-22 production in Th22 cells is controlled by the cooperative binding of Runx1 and RORγt to CNS-32 identified in this paper as the Il22 enhancer.

Recent studies have identified several transcription factors involved in the regulation of Il22 expression, but most of them have focused on its promoter (13, 14, 19, 22). CNS-32 contains conserved binding motifs for several transcription factors such as NF-κB, RORγt, Gfi1, Ets, Runx, and AP-1. Among these factors, the AP-1 family member BATF plays an important role in Th17 differentiation (47). AP-1 is a heterodimeric transcription factor composed of the Jun and Fos family members (48). In Th17 cells, BATF preferentially forms a heterodimer with JunB and binds to the promoters of key Th17-related genes, including Il17a, Il21, Il22, Rorc, Rora, and Ahr (22). T cells from Batf-deficient mice have a defective expression of the master transcription factor RORγt, which is crucial for Il17 expression and Th17 differentiation, and therefore, their mice are resistant to experimental autoimmune encephalomyelitis (22). We found that mutation of two AP-1 binding sites in CNS-32 significantly diminished the enhancer activity (Fig. 2B). However, overexpression of BATF had no effect on the enhancer activity in the reporter assays (Fig. 3A). As shown in Th17 cells, BATF might govern the initial chromatin accessibility that allows additional transcription factors such as Runx1 and RORγt to bind to and transactivate Il22, suggesting a possible pioneer role in Th22 cells.

We have mainly identified two important transcription factors, Runx1 and RORγt, which functionally synergize to augment Il22 enhancer activity by directly binding to CNS-32. Runx1 is, to our knowledge, the first transcription factor identified to be involved in the transcriptional regulation of Il22. Importantly, the Runx site is located just downstream of the RORγt site in CNS-32, which has been similarly shown to be functional in Th17 cells (40). Runx1 influences Th17 differentiation by directly inducing Rorc expression as well as interacting cooperatively with RORγt to facilitate Il17 expression through binding to its promoter and enhancer. Similar to observations in Th17 cells, overexpression of Runx1 in Th22 cells significantly induced Rorc expression and consequently led to the enhancement of IL-22 production (Fig. 7). We also found that overexpression of Runx1 and RORγt synergistically promoted IL-22 production in Th22 cells (Fig. 8). In contrast, knockdown of CBFβ with shRNA in Th22 cells resulted in much less IL-22 production (Fig. 9). Thus, the relationship between Runx1 and RORγt may represent a new axis of regulation of Il22 expression and Th22 differentiation.

Runx1 also controls Treg cell function by cooperatively binding to the master transcription factor Foxp3 (39). Foxp3, by binding to Runx1, inhibits Il17 expression mediated by the Runx1 and RORγt complex (40). This tripartite interaction among Runx1, RORγt, and Foxp3 has been proposed as a key regulatory mechanism responsible for a dichotomy in the generation of Th17 and Treg cells. In addition, the Th1-specific transcription factor T-bet also binds to Runx1 to prevent productive association of Runx1 with RORγt and consequently inhibits Rorc and Il17a expression and Th17 differentiation (49). Thus, all master transcription factors, namely RORγt, Foxp3, and T-bet, physically interact with Runx1 to modulate the generation and maintenance of specific T cell subsets.

The process of T cell differentiation may be tightly associated with stable epigenetic modifications in T cell–specific locus, and therefore, we investigated the levels of histone acetylation and DNA methylation in the Il22 locus during differentiation of naive T cells to Th22 cells. In naive T cells, histone H4 molecules were markedly hypoacetylated in the entire Il22 locus, and the CpG sites in the promoter were partially methylated, correlating with low levels of Il22 mRNA expression in these cells. However, we found that DNA demethylation has already occurred in the CNS-32 region, suggesting a high degree of accessibility to trans-acting factors in naive T cells when other cis-regulatory elements in the Il22 locus may be relatively inaccessible. Recent studies demonstrated that BATF and IFN regulatory factor 4 (IRF4) exhibit pioneer-like functions during Th17 differentiation and regulate the accessibility of transcriptional regulators to lineage-specific genomic elements (50). It is therefore likely that BATF might bind to CNS-32 in naive T cells and act as a pioneering factor. Following Th22 differentiation, Runx1 and RORγt bind to CNS-32, and histone H4 molecules become highly acetylated in the entire Il22 locus. Runx1 often influences gene transcription by acting synergistically with coactivators and histone acetyltransferases, including p300/CBP, MOZ, and MORF (51). Therefore, Runx1, in cooperation with RORγt, may exploit the enhancer landscape created by BATF to control Il22 expression by the seeding and spreading of chromatin remodeling across the entire Il22 locus.

To our knowledge, this is the first time the long-range regulation of Il22 transcription has been reported. The precise regulation of this process requires the binding of transcription factors to their cognate cis elements and subsequent changes in higher order chromatin structure. Indeed, it has been reported that several transcription factors affect the establishment of local active chromatin conformation via DNA loop formation (31, 52, 53). Therefore, we performed the ChIP and 3C assays and explored how the distant cis elements collaborate with each other for the precise regulation of Il22 transcription. When naive CD4⁺ T cells were differentiated into IL-22–producing Th22 cells, we observed the recruitment of Runx1 and RORγt to the CNS-32 enhancer and subsequent direct interaction of CNS-32 with the Il22 promoter (Fig. 6). In contrast, this long-range DNA interaction was not observed in Th1 or MEF cells. Because no significant Runx1 or RORγt binding was detected at the Il22 promoter, the lineage-specific looping may result in a locally high concentration of Runx1, RORγt, and chromatin-modifying factors near the promoter and activate Il22 transcription.

Proper regulation of gene transcription requires that genes be protected by chromatin insulators from undesired effects of regulatory elements closely linked to nearby genes. We have previously identified three CTCF binding elements, one upstream of Ifng, one within Ifng, and one downstream of Ifng, that appear to define limits of the Ifng locus (31). These elements have insulator activities and cooperate with one another to approximate interposed enhancers to the Ifng promoter region via DNA looping in a Th1-specific and CTCF-dependent manner. This looping segregates Ifng from the upstream Il22 and from other sequences downstream on the same locus. CNS-32 identified in this paper as the Il22 enhancer is located 32 and 270 kb upstream of the Il22 and Ifng promoters, respectively (Supplemental Fig. 1A). Interestingly, CNS-32 affected not only Il22 promoter activity but also Ifng promoter activity in reporter assays (Fig. 1D). These cytokine genes are typically expressed by different Th cell subsets (Supplemental Fig. 2A). IFN-γ expression is observed in Th1 cells, whereas IL-22 expression is predominantly in Th17 and Th22 rather than Th1 cells. Owing to their physical proximity, differential expression of Ifng and Il22 may require a chromatin insulator. Because the upstream CTCF element (CTCF-70) is located between Il22 and Ifng, it may function as a chromatin insulator to separate the transcriptional units of these two genes, thereby preventing promiscuous Ifng transcription in Th22 cells by the CNS-32 enhancer. Further studies will be needed to determine whether DNA looping promotes competition between Ifng and Il22 for shared regulatory elements or facilitates their coordinate expression.

In summary, we have demonstrated an important role for a novel enhancer CNS-32, which upregulates Il22 expression mediated by synergistic action of Runx1 and RORγt. Mechanistically, CNS-32 may have a functional role in shaping chromatin structure in differentiating Th22 cells as well as facilitating access of transcription factors such as Runx1 and RORγt to the Il22 locus. Our study may provide insight into the regulatory mechanism of Th22 differentiation and the pathogenesis of Th22 cell–mediated inflammatory diseases.

We thank Dr. T. Kitamura (The University of Tokyo) for providing the Plat-E cells. We thank the late Y. Sekimata for continuous support and discussion. We thank Y. Suina (Yamagata University) for technical assistance.

The authors have no financial conflicts of interest.