STAT6-mediated displacement of polycomb by trithorax complex establishes long-term maintenance of GATA3 expression in T helper type 2 cells

The expression of GATA3 mRNA is regulated in a tissue-specific manner. Naive CD4 T cells express a moderate level of GATA3, and in vitro differentiated Th2 cells and fully developed Th2 cells express higher levels of GATA3, whereas splenic B cells express little GATA3 mRNA (Fig. 1 A, left). A similar tissue-specific profile in the protein expression of GATA3 mRNA was observed (Fig. 1 A, right). Fully developed Th2 cells were established as described in Materials and methods. A schematic representation of the GATA3 gene locus, with the location of specific primer pairs and probes for quantitative PCR used in this study, is shown in Fig. 1 B. First, the histone modification and the binding of the PcG and TrxG complexes at the GATA3 gene locus were determined by chromatin immunoprecipitation (ChIP) assays. In B cells, Bmi1 bound to the upstream region of the proximal promoter and the region around exon 1 and exon 2 (primers #1 to #11; Fig. 1 C, dark blue line). The histone H3-K27 trimethylation (H3-K27Me3) signals were enriched at a broader but similar region (primers #1 to #15). Only weak signals for Menin and MLL binding and H3-K4Me3 were detected in B cells. However, in naive CD4 T cells, the Bmi1 signal was enriched only in the upstream region of the proximal promoter (primers #1 to #7; Fig. 1 C, turquoise line) but was low at the proximal promoter and the coding regions (primers #7 to #20). The signal pattern for H3-K27Me3 was almost exactly the same as that of Bmi1. In sharp contrast, Menin and MLL binding and H3-K4Me3 signals were enriched from the downstream region of the proximal promoter to exon 2. Fully developed Th2 cells showed strong enrichment of Menin and MLL signals from the beginning of the distal promoter to exon 5 and H3-K4Me3 signals to exon 3 (Fig. 1 C, red line). No significant Bmi1 accumulation and H3-K27Me3 were detected in fully developed Th2 cells. The real PCR product bands are shown in Fig. S1 (A and B). We detected comparable levels of histone H3 at all regions in these three types of cells (Fig. S1, C and D). The signals for H3-K4Me3 showed a similar pattern to that of Menin and MLL in all cells, whereas the H3-K27Me3 profile was similar to that of Bmi1. These results indicate that the increase in the expression of GATA3 mRNA correlates with the dissociation of Bmi1 and the association of Menin and MLL to the GATA3 gene locus, particularly in the upstream region of the proximal promoter (primers #3 to #6).

Next, the change of chromatin status and the PcG/TrxG binding at the GATA3 gene locus during Th2 cell differentiation were assessed using a quantitative PCR assay system (Fig. 1 D). PcG proteins, Bmi1, EZH2, and Suz12 bound to the upstream region of GATA3 proximal promoter in naive CD4 T cells (Fig. 1 D, left, probes (−15), (−5), and (−4)). The signal for H3-K27Me3 was also enriched in the upstream region of the proximal promoter. A progressive dissociation of Bmi1, EZH2, and Suz12 and decreased H3-K27Me3 were observed during Th2 cell differentiation. Essentially no PcG association was observed in fully developed Th2 cells (Fig. 1 D, left, probes (−15), (−5), and (−4)). In contrast, the binding of Menin, MLL, and RNA polymerase II (RNAPII) was restricted to the coding region in naive CD4 T cells (Fig. 1 D, right). H3-K4Me3 and H3-K9Ac signals were enriched around the proximal promoter (probe (0)). This accumulation of Menin, MLL, and RNAPII was observed in developing Th2 cells accompanied by increased H3-K4Me3 and H3-K9Ac signals across the GATA3 locus (Fig. 1 D, right, probes (−5) to (+3)). The accumulation was more prominent in fully developed Th2 cells. The total H3 levels were similar in these cells (Fig. 1 D, bottom left). Thus, the displacement of the PcG complex by the TrxG complex occurred during Th2 cell differentiation. Interestingly, the displacement was most characteristic in the region between the GATA3 proximal and distal promoter (Fig. 1 D, probes (−5) and (−4)). A previous study showed that the reduced expression of the PcG gene caused the dissociation of PcG molecules from the target genes in the epidermal progenitors (Ezhkova et al., 2009). However, in this study, substantial levels of PcG and TrxG mRNA and protein were expressed in naive CD4 T cells, Th2 cells, and fully developed Th2 cells (Fig. S2), suggesting that the displacement is unlikely to be explained by the expression levels of these molecules.

We assessed the role of the IL-4–STAT6 signaling pathway to help elucidate the molecular mechanisms underlying the displacement of the PcG complex by the TrxG complex at the GATA3 gene locus. Although naive CD4 T cells from STAT6-deficient mice expressed equivalent levels of GATA3 mRNA, they failed to up-regulate GATA3 mRNA during in vitro differentiation into Th2 cells (Fig. 2, A and B). The reduced expression of GATA3 protein in STAT6-deficient Th2 cells was confirmed by the intracellular staining of GATA3 (Fig. 2 C). WT and STAT6-deficient CD4 T cells were cultured under Th2 conditions for 5 d, and then the cells were subjected to a ChIP assay with quantitative PCR. In the case of WT and STAT6-deficient naive CD4 T cells, equivalent binding of Bmi1, Menin, and MLL and equivalent histone modifications at the upstream region and coding region of the GATA3 gene (probes (−5) and (−4)) were observed (Fig. 2 D). In contrast, the binding of Bmi1 and the level of H3-K27Me3 at the upstream region of the proximal promoter (probes (−5) and (−4)) were reduced in WT but not in STAT6–deficient effector Th2 cells (Fig. 2 D, left). In addition, the recruitment of Menin and MLL and the induction of H3-K4Me3 and H3-K9Ac were significantly impaired in STAT6-deficient Th2 cells (Fig. 2 D, right). Although less characteristic, a similar pattern was observed at the coding region (Fig. 2 D, probe (+3)). The total H3 levels were not affected in STAT6-deficient naive and Th2 cells. The displacement of the PcG complex by the TrxG complex and corresponding histone modifications observed in Th2 cells were not observed in Th1 cells (Fig. 2 E). These results indicate that the activation of the STAT6 is crucial for the displacement of the PcG complex by the TrxG complex and that this is accompanied by the alteration in histone modification of the GATA3 gene locus during Th2 cell differentiation.

A sequence analysis identified 11 putative STAT6 consensus sites within the GATA3 locus (from 20 kb upstream to 20 kb downstream relative to the GATA3 translational start site), and on this basis corresponding primer pairs (S1–S11) were prepared (Fig. 3 A). Splenic CD4 T cells were stimulated with IL-4 for 1 or 24 h, after which these stimulated cells were subjected to a ChIP assay with an anti-STAT6 antibody. Apparent binding was detected by the S4 and S7 primer pairs (Fig. 3 B). STAT6 binding was observed 1 h after stimulation and maintained for at least 24 h. A ChIP assay and quantitative PCR using CD4 T cells cultured under Th2 conditions for 1 h provided quantitative confirmation of the binding of STAT6 to the GATA3 S4 and S7 regions (Fig. 3 C). In splenic B cells, phosphorylation and nuclear translocation of STAT6 were induced by IL-4, but STAT6 binding at the GATA3 gene locus was not detected (unpublished data). The putative STAT6-binding sites are located at positions 1.5 kb (S4) and 10.2 kb (S7) relative to the translational start site (Fig. 3 A). To determine whether these motifs are required for the binding of STAT6, an electrophoretic mobility shift assay (EMSA) was used with nuclear extracts prepared from IL-4–stimulated T cells and incubated with radiolabeled double-stranded DNA oligonucleotide probes containing the putative STAT6-binding sites or mutated STAT6-binding sequence (probes S4, Mut-S4, S7, and Mut-S7). IL-4 treatment resulted in the formation of a nucleoprotein complex at these sites in a STAT6-dependent manner for probes S4 and S7 (Fig. 3 D) but no specific bands for the mutated oligonucleotides (Fig. 3 D, probes Mut-S4 and Mut-S7). Next, a supershift EMSA was performed using WT CD4 T cells stimulated in the presence of IL-4 (Fig. 3 E). Whereas no band shift was observed with control antibodies (Fig. 3, compare D [second through eighth lanes] with E [first through fifth lanes]), bands were supershifted by preincubation with antibodies against STAT6 (Fig. 3 E, fourth through eighth lanes), indicating that these complexes contain STAT6. Antibodies specific for HAT complex molecules, anti-p300, or anti–CREB-binding protein (CBP) mAb were also included in the supershift analysis. The intensity of the STAT6 complex band was reduced by the addition of anti-p300 mAb (second and sixth lanes) and anti-CBP mAb (Fig. 3 E, third and seventh lanes), indicating that p300 and CBP are components in the STAT6 complex. Thus, STAT6 appears to associate with the p300/CBP-containing HAT complex after IL-4 stimulation.

To further understand the role for histone modification in this process, the effect of histone deacetylase inhibitor TSA in the displacement of PcG complex by TrxG complex was examined. WT and STAT6-deficient CD4 T cells were cultured under Th2 conditions in the absence or presence of TSA for 5 d, and then the cells were subjected to a ChIP assay. The dissociation of the PcG complex (Bmi1 and EZH2) at the upstream region of the proximal promoter (probes (−5) and (−4)) occurred normally in STAT6-deficient Th2 cells in the presence of TSA (Fig. 4 A, top). H3-K27Me3 at the GATA3 locus was also reduced in the presence of TSA. The total H3 levels were not influenced by TSA treatment. In contrast to the dissociation of the PcG complex, no obvious recruitment of Menin or MLL was detected in the TSA-treated STAT6-deficient Th2 cells (Fig. 4 B, top). Interestingly, TSA treatment induced H3-K4Me3 in addition to the H3-K9Ac at the GATA3 locus (Fig. 4 B, bottom). Next, the effect of TSA on the GATA3 transcription was examined. The transcripts between the distal and the proximal promoter (probes (−10) to (−4)) were not induced in STAT6-deficient Th2 cells, and these transcripts were induced by the treatment with TSA (Fig. 4 C, top left). However, the TSA treatment did not induce mature GATA3 transcript (Fig. 4 C, top right).

Recent studies indicate that the phosphorylation status at Ser2 and Ser5 is associated with the transcriptional active and poised status of the RNAPII complex (for review see Brookes and Pombo, 2009). Consequently, we performed ChIP assays to assess the binding of Ser2- and Ser5-phosphorylated RNAPII at the GATA3 gene locus. No obvious difference in the binding pattern was observed in both WT and STAT6-deficient Th2 cells (Fig. S3). The TSA treatment resulted in the accumulation of RNAPII (both Ser2- and Ser5-phosphorylated RNAPII) between the distal and the proximal promoter (Fig. 4 C, bottom left, probes (−10) to (−4)) but not at the coding region of the proximal promoter (Fig. 4 C, bottom right, probe (+3)). No obvious decrease but rather a slightly increased expression of Bmi1, EZH2, Menin, and MLL in the STAT6-deficient Th2 cells was seen with TSA treatment, suggesting that neither PcG complex dissociation nor TrxG complex recruitment in the TSA-treated Th2 cells is controlled by the expression levels of these molecules (Fig. S4). Thus, histone hyperacetylation itself is responsible for the dissociation of the PcG complex accompanied with down-regulation of H3-K27Me3 and noncoding transcripts at the upstream region of the proximal promoter of the GATA3 gene; however, this alone was not sufficient for the recruitment of the TrxG complex.

Next, we assessed whether STAT6 activation itself induced the recruitment of Menin and mature transcripts. STAT6VT, an autoactivated STAT6 molecule (Daniel et al., 2000), and a STAT6VT DNA-binding mutant (STAT6VT dDBD) molecule (Fig. 5 A) were introduced into STAT6-deficient CD4 T cells cultured under Th2 conditions. The STAT6VT- and STAT6VT dDBD–introduced cells were harvested 3 d after infection and then were subjected to a ChIP assay. STAT6VT was found to induce the recruitment of Menin between the distal and the proximal promoter and also the coding region (Fig. 5 B, probes (−5), (−4), and (+3)). STAT6VT induced the reduction of H3-K27Me3 and the up-regulation of H3-K4Me3 (Fig. 5 C). However, STAT6VT dDBD did not induce the recruitment of Menin, the reduction of H3-K27Me3, or the up-regulation of H3-K4Me3 (Fig. 5, B and C). The total H3 levels were not influenced by STAT6VT introduction (Fig. 5 C). In addition, the transcripts between the distal and the proximal promoter (probes (−10) to (−4)) were restored by the ectopic expression of STAT6VT in STAT6-deficient developing Th2 cells (Fig. 5 D, left). Furthermore, STAT6VT rescued the mature GATA3 transcript (Fig. 5 D, right), whereas STAT6VT dDBD had no effect on GATA3 transcription (Fig. 5 D). These results indicate that the activation of STAT6 itself induced the recruitment of Menin and the mature GATA3 transcript during the development from naive CD4 T cells to Th2 cells.

The first alternative exons of the GATA3 transcripts (1a and 1b) are spliced to a common exon 1, which contains the translation start site (Fig. 5 E; Asnagli et al., 2002; Scheinman and Avni, 2009). First, the levels of transcripts of both exon 1a-1 and exon 1b-1 were assessed in WT and STAT6-deficient Th2 cells in the presence of TSA (Fig. 5 F). We confirmed that the exon 1b-1 transcript was much more abundant than that of exon 1a-1 (Scheinman and Avni, 2009). The levels of both transcripts were very low in STAT6-deficient Th2 cells. Interestingly, TSA treatment partially rescued the exon 1a-1 transcript but not the exon 1b-1 transcript (Fig. 5 F). The result was similar to that seen with noncoding and mature transcripts, respectively, as shown in Fig. 4 C. Then, we assessed the effect of the ectopic expression of STAT6VT on these two transcripts and observed that rescue was partial for the exon 1a-1 transcript but was substantial in the case of the exon 1b-1 transcript (Fig. 5 G). Thus, the effect of STAT6VT on exon 1b-1 transcript was similar to that seen on GATA3 mature transcripts, as shown in Fig. 5 D. STAT6VT dDBD showed no obvious effect on exon 1a-1 and exon 1b-1 transcripts (Fig. 5 G). These results indicate that both exon 1a-1 and exon 1b-1 transcripts are dependent on STAT6 activation and that the former is associated with noncoding transcripts of the upstream region of the GATA3 gene and the latter is associated with the mature transcripts.

Menin is an essential component for DNA binding of the TrxG–MLL complex (Guru et al., 1998). A CD4-Cre–driven conditional knockout mouse for Menin (Menin-deficient mice) was used to assess the role of Menin in the TrxG recruitment in the expression of the GATA3 gene. In Menin-deficient mice, equivalent numbers of CD4 and CD8 T cells developed in the thymus, but a 50% reduction in the number of splenic CD4 cells was observed in comparison with WT control (Fig. S5 A). The cell surface phenotype of splenic CD4 T cells was found to be within the normal range (Fig. S5 B). Proliferative responses induced with immobilized anti-TCR mAb were not impaired but somewhat slightly enhanced in Menin-deficient CD4 T cells (unpublished data). The lack of Menin mRNA in the conditional Menin-deficient Th1 and Th2 cells was confirmed (Fig. S5 C). Menin-deficient CD4 T cells cultured under Th2 conditions expressed an equivalent amount of GATA3 transcript and protein to WT CD4 T cells (after the first cycle of stimulation; Fig. 6, A and B) and differentiated normally into effector Th2 cells (Fig. 6 D). To further probe the role of Menin, the binding of MLL and the status of H3-K4Me3 at the GATA3 gene locus were assessed in the Menin-deficient Th2 cells. We could not detect the recruitment of MLL at the GATA3 gene locus in Menin-deficient Th2 cells (Fig. 6 C, left). Interestingly, the levels of H3-K4Me3 were decreased at the upstream region of the GATA3 proximal promoter (probes (−5) and (−4)), whereas this was not obvious at the coding region (Fig. 6 C, left, probe (+3)). The binding of the PcG proteins, Bmi1 and EZH2, and those of H3-K27Me3 after the first cycle stimulation were reduced equivalently between the WT and Menin-deficient Th2 cells (Fig. 6 C, right). The total H3 levels were not changed in Menin-deficient Th2 cells (Fig. S6 A). These results indicate that STAT6-dependent but Menin–TrxG-independent GATA3 transcription is induced during the first cycle of Th2 cell differentiation.

Interestingly, although the expression levels of GATA3 transcript in WT Th2 cells was maintained even when the cells were stimulated with anti-TCR mAb in the presence of anti–IL-4 mAb (Fig. S7, second to third cycle), the expression of GATA3 transcript in Menin-deficient Th2 cells was not maintained in the presence of anti–IL-4 mAb (after the third cycle of stimulation; Fig. 6 E). The high expression of GATA3 protein in Menin-deficient Th2 cells was also not maintained (Fig. 6 F). The recruitment of MLL at the GATA3 gene locus was not detected in Menin-deficient Th2 cells after third cycle cultivation (Fig. 6 G, left). H3-K4Me3 and H3-K9Ac levels were decreased at both the upstream and the coding regions after stimulation in the presence of anti–IL-4 mAb (Fig. 6 G, left). Interestingly, no significant increase in the binding of Bmi1 and EZH2 and in the level of H3-K27Me3 was observed even after stimulation in the presence of anti–IL-4 mAb (Fig. 6 G, right). The total H3 levels were not changed in Menin-deficient Th2 cells even after the third cycle cultivation (Fig. S6 B). These results indicate that the PcG complex did not reassociate with the GATA3 gene locus in Th2 cells even in the absence of the binding of the Menin–TrxG complex. The ability to produce IL-4 was assessed after restimulation with anti-TCR mAb and was found to be diminished in the Menin-deficient Th2 cells after the two-cycle stimulation in the presence of anti–IL-4 mAb (Fig. 6 H). The antigenic stimulation with IL-4 for 1 wk did not restore the production of IL-4 in the Th2 cells that had received two-cycle stimulation in the presence of anti–IL-4 mAb (unpublished data), indicating that the ability to become IL-4–producing cells was lost in the absence of Menin. Menin-deficient naive CD4 T cells expressed an equivalent amount of GATA3 transcript to WT CD4 T cells (Fig. 6 I, 0 cycle). The expression of GATA3 transcript in Menin-deficient Th2 cells was not maintained in the presence of anti–IL-4 mAb (Fig. 6 I, second cycle). Collectively, once Th2 cells have successfully developed, they maintain the GATA3 expression and Th2 function via the recruitment of the Menin–TrxG complex to the GATA3 gene locus even in the absence of IL-4–mediated STAT6 activation.

This study dissects two distinct processes that are critical in the regulation of transcription of the GATA3 gene in Th2 cells based on the binding of PcG and TrxG complexes (Fig. 7). In naive CD4 T cells, the PcG complex binds to the upstream region of the GATA3 proximal promoter, whereas the accumulation of the Menin–TrxG complex is restricted to a part of the coding region. In developing Th2 cells, IL-4–mediated STAT6 activation induces the dissociation of the PcG complex, which is accompanied by the acetylation of H3-K9 and noncoding transcripts at the upstream region of the GATA3 gene locus. Histone hyperacetylation is sufficient to cause the dissociation of the PcG complex and the induction of noncoding transcripts. High-level transcription of the GATA3 gene is induced in these cells (Fig. 7, middle). The role of binding of the Menin–TrxG complex in GATA3 transcription appears to be minimal at this stage. In fully developed Th2 cells, high-level expression of GATA3 can be maintained in the absence of IL-4 (Fig. 7, bottom). IL-4/STAT6-independent but Menin–TrxG-dependent regulation of GATA3 transcription is operating. A broad range binding of Menin was confirmed by a ChIP-seq assay (Fig. S8). The binding of Menin–TrxG complex to the GATA3 gene is required for the long-term maintenance of GATA3 expression and Th2 cytokine production. Because the supply of IL-4 is limiting in the body after the clearance of infectious organisms, this could be the mechanism by which memory Th2 cells can maintain Th2 function for a long time in vivo.

STAT6 plays several distinct roles in the regulation of GATA3 transcription. Although STAT6 is well known to be required for the expression of the GATA3 gene, STAT6-binding sites that are critical for the induction of GATA3 transcription have not been identified. We identified two STAT6-binding sites within the intron 1 and intron 3 regions of the GATA3 gene locus at which STAT6 rapidly accumulated after IL-4 stimulation, resulting in the recruitment of the HAT complex (Fig. 3). Therefore, Th2-specific GATA3 transcription appears to be mediated by the direct binding of STAT6. These binding sites are located several kilobases downstream from the transcriptional start site. However, this is not unusual because there are several examples in which transcription is induced by the binding of STAT family proteins at the distal locus control region (LCR; Jothi et al., 2008; Liao et al., 2008).

We also demonstrate that STAT6 binding recruited the HAT complex to the GATA3 gene and induced the spreading of histone hyperacetylation with the resultant dissociation of the PcG complex at the upstream region of the GATA3 gene. The two STAT6-binding sites are also located several kilobases downstream from the region where the displacement of PcG by TrxG occurs. Two likely mechanisms can be entertained. First, these STAT6-binding sites may act as cis-regulatory elements and exert enhancer activity for GATA3 mRNA expression. At the β-globin gene locus, communication between the LCR and the active β-globin genes is generally thought to occur via a looping mechanism whereby the LCR and the β-globin genes come into direct physical contact through their interacting proteins and the intervening DNA sequences are looped out (Dean, 2006). The STAT6-binding sites may be close to the proximal or distal promoter through the high-ordered chromatin configuration, and thus, the STAT6 complex may control the dissociation of the PcG complex from the distant region of the STAT6-binding sites. Alternatively, the STAT6–HAT complex may spread from its binding sites toward the upstream region during Th2 cell differentiation and induce the dissociation of the PcG complex. This model is called the tracking or spreading model (Bondarenko et al., 2003; Dean, 2006). For example, erythroid activators NF-E2 and GATA1 are important for the recruitment of the HAT complex to the LCR and for histone acetylation at the LCR and the distant β-globin gene (Johnson et al., 2001). The accumulation of STAT6 was observed at only two sites at the early phase of Th2 cell differentiation (Fig. 3, B and C), whereas a few days after the initial stimulation, the binding of STAT6 was broadly detected throughout the GATA3 gene by a ChIP assay (unpublished data). Thus, we would favor the second model.

The induction of Th2-specific up-regulation of GATA3 transcription has been suggested to result from the activation of the distal promoter of GATA3 (Amsen et al., 2007; Fang et al., 2007). However, recent studies by other investigators have indicated that the Th2-specific increase of GATA3 transcripts was dependent on the proximal promoter (Scheinman and Avni, 2009; Yu et al., 2009). As shown in Fig. 5 (F and G), we demonstrated that exon 1b-1 transcripts induced by the proximal promoter were much more abundant in comparison with exon 1a-1 transcript, and thus, it would appear to us that the majority of the Th2-specific increase of mature GATA3 transcripts is dependent on the proximal promoter. In contrast, the transcript induced by the distal promoter may reflect the existence of noncoding transcripts, which may play an important role in the dissociation of the PcG complex at the GATA3 gene (Hekimoglu and Ringrose, 2009). In fact, the induction of noncoding transcripts and that of exon 1a-1 junction-containing transcripts occurs in similar fashion. The induction of noncoding transcripts at the upstream region of the proximal promoter including the region of the distal promoter was STAT6 dependent, and the noncoding transcripts were induced by the treatment with TSA (Fig. 4 C). The transcripts from the distal promoter are spliced and form the exon 1a-1 junction (Fig. 5 E, middle). These exon 1a-1 junction-containing transcripts were also STAT6 dependent and significantly up-regulated by the treatment with TSA (Fig. 5 F). The polycomb response element (PRE) and the trithorax response element (TRE) have been well established in Drosophila (Schwartz and Pirrotta, 2008). Noncoding transcripts have been detected around the PRE/TRE, and they play an important role for the binding of the PcG and TrxG complexes (Schmitt et al., 2005; Hekimoglu and Ringrose, 2009). Therefore, the identification of the PRE/TRE in the region between exon 1a and exon 1b of the GATA3 gene locus would be an interesting issue.

In summary, this study provides the first evidence indicating that the activation of the transcription factor STAT6 induces the displacement of the PcG complex by the TrxG complex, which then leads to control of the expression of the GATA3 gene in Th2 cells. STAT6-induced histone acetylation appeared to be responsible for the dissociation of the PcG complex. The recruitment of the Menin–TrxG complex was found to be crucial for the maintenance of high GATA3 expression in developed Th2 cells. Therefore, STAT6-dependent displacement of the PcG by the TrxG complex would allow for a permissive chromatin status of the GATA3 gene locus in the developed Th2 cells in which a STAT6-independent regulation of transcription operates.

C57BL/6 and BALB/c mice were purchased from CLEA. STAT6-deficient mice (Takeda et al., 1996) were provided by S. Akira (Osaka University, Suita, Osaka, Japan). Menin-deficient mice (Crabtree et al., 2003) were purchased from The Jackson Laboratory. CD4-Cre transgenic mice were purchased from Taconic. All mice used in this study were maintained under specific pathogen-free conditions and ranged from 6 to 8 wk of age. All experimental protocols using mice were approved by the Chiba University animal committee. All animal care was performed in accordance with the guidelines of Chiba University.

Recombinant mouse IL-12 was purchased from BD, and recombinant mouse IL-4 was purchased from TOYOBO. The OVA peptide (residues 323–339; ISQAVHAAHAEINEAGR) was synthesized by BEX Corporation.

The antibodies used for the ChIP assay were anti–acetyl-histone H3-K9 (Millipore), anti–trimethyl-histone H3-K4 (LP Bio), anti–trimethyl-histone H3-K27 (Millipore), anti–histone H3 (Abcam), anti–Ser5-P RNAPII (Abcam), anti–Ser2-P RNAPII (Abcam), anti-MLL (Bethyl Laboratories, Inc.), anti-Bmi1 (Santa Cruz Biotechnology, Inc.), anti-STAT6 (Santa Cruz Biotechnology, Inc.), anti-Menin (Bethyl Laboratories, Inc.), anti-Suz12 (Abcam), and anti-EZH2 (Diagenode). The antibodies used for EMSA were anti-STAT6 (Santa Cruz Biotechnology, Inc.), anti-CBP (Abcam), and anti-p300 (Millipore). The antibodies used for cytoplasmic staining were anti–IFN-γ–FITC, anti–IL-4–PE, and anti-GATA3–Alexa Fluor 647 (BD).

Th1/Th2 cells were generated as previously described (Yamashita et al., 2006). In brief, splenic CD4 T cells were stimulated with 3 µg/ml of immobilized anti–TCR-β mAb plus 1 µg/ml anti-CD28 mAb under the Th1 or Th2 culture conditions for 5 d in vitro. Th1 conditions were as follows: 25 U/ml IL-2, 10 U/ml IL-12, and anti–IL-4 mAb. Th2 conditions were as follows: 25 U/ml IL-2 and 100 U/ml IL-4. These cells were used as either Th1 or Th2 cells.

Splenic CD4 T cells from DO11.10 OVA-specific TCR transgenic mice were stimulated with an OVA peptide (1 µM Loh15) plus APC under Th2 culture conditions for 5 d in vitro. The Th2 cells were further cultured in vitro for 2 d in the absence of any exogenous cytokines. The cultured CD4 T cells were then restimulated with OVA peptide (1 µM Loh15) plus APC with IL-2 and anti–IL-4 mAb for 5 d. This cycle was then repeated more than three times.

Total RNA was isolated using the TRIZOL reagent (Invitrogen). cDNA was synthesized using oligo (dT) primer and Superscript II RT (Invitrogen). Quantitative RT-PCR was performed as described previously using a sequence detection system (ABI Prism 7500; Applied Biosystems; Yamashita et al., 2006). The primers and TaqMan probes for the detection of GATA3, Bmi1, EZH2, Menin, MLL, and HPRT were purchased from Applied Biosystems and Roche, respectively. The specific primers and Roche Universal probes used are described in Table S1. The expression was normalized by the HPRT signal.

ChIP was performed using ChIP assay kits (Millipore) as previously described (Yamashita et al., 2006). Quantitative representations of the results are shown as relative band intensities measured by a densitometer (AE6905H [ATTO] and CS Analyzer version 2.08b). The specific primers used are described in Table S1. Real-time quantitative PCR analysis was performed on an ABI Prism 7500 real time PCR machine with TaqMan probes and primers (sequences available in Table S1). To calculate the enrichment of each protein to a particular target DNA, values obtained (via the standard curve method) for each target were divided by the amount of the corresponding target in the input fraction. Enrichments obtained from mock immunoprecipitations performed in parallel with normal IgG were then subtracted from the enrichment values obtained with specific antibodies ([specific antibody ChIP − control Ig ChIP]/input DNA). All the enrichments are expressed as a function of the highest enrichment obtained on the locus (set to 10; Demers et al., 2007).

EMSAs were performed using a gel shift assay system (Promega) as described previously (Kimura et al., 2005). In brief, the nuclear extracts were incubated at room temperature with a ³²P-labeled, double-stranded oligonucleotide in DNA-binding buffer. In some experiments, the nuclear extracts were preincubated at 4°C with specific antibodies. Electrophoresis was conducted on 4% native polyacrylamide gel (acrylamide/bisacrylamide ratio 29:0.8 in 0.5× Tris-borate-EDTA), and the radioactivity was visualized by autoradiography. The oligonucleotides used in this experiment were normal probe S4, 5′-CTTGGCGTTCAAGAGAATTCTCAA-3′; mutant probe S4, 5′-CTTGGCGTTCGGTTTAATTCTCAA-3′; normal probe S7, 5′-AGCCAACTTCCTAGGAAAAAGCTG-3′; and mutant probe S7, 5′-AGCCAACTTCAGTTTAAAAAGCTG-3′. The STAT6 consensus motif is underlined, and the mutated nucleic acids are shown in italicized and bold characters.

STAT6-deficient splenic CD4 cells were cultured under Th2 conditions, and 10 nM TSA (Sigma-Aldrich) was added in the culture on day 2. After another 3-d culture, CD4 T cells were collected for the ChIP assay and RT-PCR.

The pMX-IRES-hNGFR plasmid was generated as previously described (Shinnakasu et al., 2006). Retrovirus vectors containing STAT6VT cDNA (pMXs-STAT6VT-IRES-hNGFR; Daniel et al., 2000) or STAT6VT dDBD cDNA (pMXs-STAT6VT-dDBD-IRES-hNGFR) were used. The infected cells were enriched by magnetic cell sorting with anti-hNGFR (clone C40-1457; BD) and were subjected to a ChIP assay.

Naive CD4 T cells from Menin-deficient mice were cultured under Th2 conditions for 5 d. The Th2 cells were further cultured for 2 d in the absence of cytokines added and then restimulated with anti-TCR mAb in the presence of IL-2 and anti–IL-4 mAb for an additional 5 d. This cycle was then repeated. PcG/TrxG binding and histone modifications were detected by ChIP assays, and the GATA3 expression was determined using quantitative RT-PCR at the end of the each cycle.

Cytoplasmic extracts and nuclear extracts were prepared using NE-PER Nuclear and Cytoplasmic Extraction Reagent (Thermo Fisher Scientific). The antibodies used for the immunoblot analysis were anti–histone H3 (Abcam), anti-Erk1 (Santa Cruz Biotechnology, Inc.), anti-MLL (Bethyl Laboratories, Inc.), anti-Bmi1 (Santa Cruz Biotechnology, Inc.), anti-Menin (Bethyl Laboratories, Inc.), and anti-EZH2 (Diagenode).

For ChIP-seq analysis, immunoprecipitate and input samples were prepared using ChIP-Seq Sample Prep kit (Illumina, Inc.). Adaptor-ligated DNA fragments were size-fractionated by 12% acrylamide gel, and the 150–250-bp fraction was recovered. DNA thus obtained was amplified by 18 cycles of PCR. 1 ng of the DNA was used for the sequencing reaction of the GAIIx (Illumina, Inc.) according to the manufacturer’s instructions. 150,000–250,000 clusters were generated per tile, and 36 cycles of the sequencing reactions were performed. Short-read sequences were aligned to the mouse genome sequences (mm9 as from the University of California, Santa Cruz Genome Browser) using the Eland program. Sequences allowing no more than two mismatches per sequence were used for the analysis.

Fig. S1 shows the real PCR product bands of the results shown in Fig. 1 C. Fig. S2 shows expression of Bmi1, EZH2, Menin, and MLL in naive CD4 T cells, Th2 cells, and fully developed Th2 cells. Fig. S3 shows the binding pattern of phosphorylated RNAPII at the GATA3 gene locus. Fig. S4 shows that expression of Bmi1, EZH2, Menin, and MLL was not affected by TSA treatment in the STAT6-deficient Th2 cells. Fig. S5 shows phenotypic characterization of peripheral CD4 T cells in Menin-deficient mice. Fig. S6 shows the levels of total H3 at the GATA3 gene locus in Menin-deficient Th2 cells. Fig. S7 shows IL-4/STAT6-independent maintenance of GATA3 expression in developed Th2 cells. Fig. S8 shows binding of Menin protein to specific regions around the GATA3 gene locus. Table S1 shows primers and probes used for ChIP and RT-PCR.

We are grateful to Drs. Ralph Kubo and Anjana Rao for their helpful comments and constructive criticisms in the preparation of the manuscript. We thank Ms. Kaoru Sugaya, Hikari K. Asou, Satoko S. Norikane, and Mr. Toshihiro Ito for their excellent technical assistance.

This work was supported by the Global Centers of Excellence Program (Global Center for Education and Research in Immune System Regulation and Treatment), Ministry of Education, Culture, Sports, Science and Technology (MEXT; Japan), and by grants from MEXT (Grants-in-Aid for Scientific Research on Priority Areas #17016010 and #20060003, Scientific Research [B] #21390147, Scientific Research [C] #21591808 and #20590485, and Young Scientists [B] # 20790367, [Start-up] #19890041, [Start-up] #20890038, and #2056131), the Ministry of Health, Labour and Welfare (Japan), and the Uehara Memorial Foundation.

The authors have no conflicting financial interests.