CD28 costimulation controls histone hyperacetylation of the interleukin 5 gene locus in developing th2 cells.

Interleukin 5 (IL-5) plays a unique role in allergic inflammatory responses, and the understanding of molecular mechanisms underlying the generation of IL-5-producing cells is crucial for the regulation of allergic disorders. Differentiation of naive CD4 T cells into type-2 helper (Th2) cells is accompanied by chromatin remodeling including hyperacetylation of histones H3 and H4 in the nucleosomes associated with the IL-4, IL-13, and IL-5 genes. Histone hyperacetylation of the IL-5 gene displayed a delayed kinetics compared with that of the IL-4 and IL-13 genes, suggesting a distinct remodeling mechanism for the IL-5-gene locus. Here we studied the role of CD28 costimulation in the generation of IL-5-producing cells and the histone hyperacetylation of the IL-5 gene locus. CD28-costimulation selectively enhanced histone hyperacetylation of the IL-5 gene locus that appeared to be mediated through NF-kappaB activation and subsequent up-regulation of GATA3. The CD28 costimulation-sensitive histone hyperacetylation spanned almost the entire intergenic region between the IL-5 and RAD50 accompanied with intergenic transcript. Thus, this is the first demonstration that CD28 costimulation controls a chromatin-remodeling process during Th2 cell differentiation.

Upon antigen recognition by T cell receptor (TCR), 1 naive CD4 T cells differentiate into two distinct helper T (Th) cell subsets, Th1 and Th2 cells (1). Th1 cells produce IFN␥ and tumor necrosis factor-␤ and initiate cell-mediated immunity against intracellular pathogens. Th2 cells produce IL-4, IL-5, and IL-13 and are involved in humoral immunity and allergic responses. The cytokine environment is crucial in controlling the direction of Th cell differentiation (2,3). For Th1 cell differentiation, IL-12-mediated activation of signal transducer and activator of transcription (STAT) 4 is required, whereas IL-4-mediated STAT6 activation is important for Th2 cell generation (4 -6). In addition, TCR stimulation events upon encounter with antigens are also indispensable for both Th1 and Th2 cell differentiation. We reported that efficient TCR-mediated activation of the p56 lck , calcineurin, and Ras-extracellular signal-regulated kinase mitogen-activated protein kinase signaling cascade is crucial for Th2 cell differentiation (7)(8)(9). Recent studies have identified several transcription factors that control Th1/Th2 cell differentiation (10). Among them, GATA3 appears to be a master transcription factor for Th2 cell differentiation. GATA3 is selectively induced in developing Th2 cells, and the ectopic expression of GATA3 induced Th2 cell differentiation even in the absence of IL-4 or STAT6 (11)(12)(13)(14). For Th1 cell differentiation, T-bet was recently identified as a key transcription factor (15).
CD28 costimulation enhances Th2 responses significantly (16,17). Upon anti-CD28 mAb stimulation, phosphatidylinositol 3-kinase is recruited to CD28 and activated, and then subsequent activation of NF-B is induced (18 -21). It has been reported that GATA3 induction was an outcome of the CD28induced NF-B activation in T cells (22,23). This may be a mechanism by which Th2 responses were enhanced by CD28 costimulation. It is also known that IL-5 production and IL-5dependent airway inflammation are dependent on NF-B family members (24 -26).
Chromatin remodeling of the Th2 cytokine gene loci (IL-4/ IL-5/IL-13) occurs during Th2 cell differentiation (27). A highly conserved 400-bp noncoding sequence 1 (CNS1) was identified, and an important role in coordinate expression of Th2 cytokines was revealed (28,29). More recently, a 3Ј distal IL-4 enhancer (V A ) containing an inducible DNase I hypersensitive site was identified (30). Reiner and co-workers (31) report that demethylation of the intron 2 region of the IL-4 gene was associated with cell cycle progression and Th2 cell differentiation (31). We reported that demethylation of this region is regulated by polycomb group genes (32) that are known to regulate transcriptional memory in Drosophila.
Hyperacetylation of histone H3 and H4 by histone acetyltransferases was suggested to be associated with active chromatin (33). Recently, we and others have demonstrated that histone hyperacetylation of the Th2 cytokine gene loci occurs in developing Th2 cells in a Th2-specific and STAT6-dependent manner (34 -36). We demonstrated an essential role for GATA3 in the Th2-specific hyperacetylation (34). We also generated a precise map of the Th2-specific histone hyperacetylation within the IL-13 and IL-4 gene loci and identified a 71-bp conserved GATA3 response element 1.6 kilobase pairs upstream of IL-13 locus exon 1. This histone hyperacetylation remodeling process could be a major target for the Th2 master transcription factor GATA3 to induce differentiation toward Th2 cells.
Histone hyperacetylation of another Th2 cytokine gene locus, IL-5, occurs in a Th2-specific STAT6-and GATA3-dependent manner with significantly different kinetics compared with that of the IL-4 and IL-13 genes (34). The direction of transcription of the IL-5 gene is opposite to that of IL-13 and IL-4. In addition, the RAD50 gene encoding a DNA repair enzyme is located between the IL-5 and IL-13 gene loci. A differential role for GATA3 in the regulation of promoter activity of the IL-5 gene from IL-4 has been suggested (37)(38)(39). These results encouraged us to explore possible novel molecular mechanisms that would govern histone hyperacetylation of the IL-5 gene locus.
In the present study we investigated histone hyperacetylation of the IL-5 gene locus in developing Th2 cells cultured with or without CD28 costimulation. A long range CD28-sensitive Th2-specific histone hyperacetylation was detected in the IL-5 and intergenic region of the IL-5 and RAD50 gene loci. The hyperacetylation was accompanied by CD28-sensitive intergenic transcripts and required high expression of GATA3. A molecular mechanism that governs Th2-specific histone hyperacetylation of the IL-5-gene associated nucleosomes will be discussed.

MATERIALS AND METHODS
Mice-C57BL/6 mice were purchased from SLC (Shizuoka, Japan). STAT6-deficient (KO) mice were kindly provided by Shizuo Akira (Osaka University, Osaka, Japan) (40). All mice used in this study were maintained under specific-pathogen-free conditions and were about 4 weeks old. Animal care was in accordance with the guidelines of Chiba University.
Cell Cultures and in Vitro T Cell Differentiation-Splenic CD4 T cells were stained with anti-CD4-fluorescein isothiocyanate and then purified using magnetic beads and an Auto-MACS sorter® (Miltenyi Biotec), yielding a purity of Ͼ98%. Enriched CD4 T cells (1.5 ϫ 10 6 ) were stimulated for 2 days with immobilized anti-TCR mAb (H57-597, 3 g/ml) and soluble anti-CD28 mAb (37.51, 3 g/ml) in the presence of IL-2 (25 units/ml), IL-12 (100 units/ml), and anti-IL-4 mAb (11B11, 25% culture supernatant) for Th1-skewed conditions. For Th2-skewed conditions, cells were stimulated with immobilized anti-TCR mAb as above but in the presence of IL-2 (25 units/ml), IL-4 (100 units/ml), and anti-IFN␥ mAb (R4.6A2, 25% culture supernatant). The cells were then transferred to new dishes and cultured for another 5 days in the presence of immobilized anti-TCR mAb, soluble anti-CD28 mAb, and the cytokines present in the initial culture. To enhance the generation of IL-5-producing cells, stimulation with anti-TCR and anti-CD28 mAbs was performed during the second culture for 5 days. This procedure is slightly different from that used in our previous report (16). Where indicated, wortmannin (Calbiochem) was added to the culture at the doses of 30 or 300 nM for the first 2 days. In vitro differentiation was then assessed by intracellular cytokine staining with anti-IL-4, anti-IL-5, anti-IL-13, and anti-IFN␥ or by ELISA as described (42).
Chromatin Immunoprecipitation (ChIP) Assay-The ChIP assay was performed using histone H3 ChIP assay kits (17-245: Upstate Biotechnology) as described (34). Anti-GATA3 Ab (H-48: Santa Cruz Biotechnology) was used for precipitation. Where indicated, GFP-positive retrovirus-infected cells were sorted by flow cytometry and subjected to ChIP assay. Several primer sequences for ChIP assay were described previously (34,42). The primer pairs newly generated are as follows: IL-5 1-F, 5Ј-Ϫ109 GATTGTTAGCAATTATTCATTTC Ϫ87 -3Ј; IL-5, 1-R, 5Ј-ϩ244 CACTGAGCTGCCTGGCGCCGT. A detailed protocol for detection of intergenic transcripts was described previously (34). The primers used are the same as those used in ChIP assay.

RESULTS
Costimulation with Anti-CD28 mAb Enhances the Generation of IL-5-and IL-13-producing Cells-The aim of this study was to clarify the molecular mechanisms that control chromatin remodeling of the IL-5 gene locus during Th2 cell differentiation. We first assessed the role for CD28 costimulation in the generation of IL-5-producing Th2 cells. Freshly prepared CD4 T cells from young adult (4 weeks) B6 mice were cultured in vitro with immobilized anti-TCR in the presence of agonistic anti-CD28 mAbs (37.51) to effect stimulation. The IL-5/IL-4 profiles of CD4 T cells cultured under Th1-or Th2-skewed conditions are depicted in Fig. 1A. As can be seen, the generation of IL-5-producing cells (both IL-5 ϩ IL-4 Ϫ and IL-5 ϩ IL-4 ϩ fractions) cultured under Th2-skewed conditions was greatly enhanced in the presence of CD28 costimulation. The increased generation of IL-4-producing cells was marginal under these culture conditions. IL-5-producing cells generated with CD28 costimulation were STAT6-dependent and not detected under Th1-skewed culture conditions. The generation of IFN␥-producing Th1 cells was moderately increased by the presence of CD28 costimulation (Fig. 1B). The levels of IL-13-producing cells were also increased by the presence of CD28 costimulation (Fig. 1C).
Concurrently, the amount of cytokines produced by developing Th2 cells cultured with CD28 costimulation was assessed by ELISA (Fig. 1, D and E). As expected, CD28 costimulation significantly enhanced the production of IL-5 and IL-13, whereas the effects on the production of IL-4 and IFN␥ were marginal. The production of Th2 cytokines (IL-5, IL-13, and IL-4) was all STAT6-dependent regardless of the presence or absence of CD28 costimulation (Fig. 1E). Taken together these results suggest that the generation of IL-5-and IL-13-producing cells was Th2-specific, STAT6-dependent, and more sensitive to CD28 costimulation as compared with that of IL-4producing cells.
Dynamics of Histone H3 Hyperacetylation of the IL-5 Gene Locus in Developing Th2 Cells Cultured with CD28 Costimulation-To clarify whether CD28 costimulation enhances histone hyperacetylation of the IL-5 gene locus during Th2 cell differentiation, we first examined the kinetics of acetylation of the IL-5 promoter, IL-4 promoter, and RAD50 promoter regions using a ChIP assay with anti-acetylhistone H3 Ab. Developing Th1 and Th2 cells cultured with CD28 costimulation were harvested 2, 3.5, 5, and 7 days after stimulation. The results of this analysis are presented as the relative band intensity of each group normalized with band intensity of the corresponding input DNA as shown in the right panel of Fig. 2. Two days after stimulation, low but significant levels of increase in acetylation occurred at all of the regions tested, and basically no difference between the three culture conditions was noted. However, as shown in Fig. 2A, the levels of histone acetylation associated with the IL-5 promoter increased significantly after cultivation for more than 3 days under Th2 skewed conditions, particularly in those cultures with CD28 costimulation. In contrast, the levels of histone hyperacetylation in the IL-4 promoter were increased day by day, and no significant difference was detected in the presence or absence of CD28 costimulation. The levels of histone hyperacetylation in the RAD50 gene promoter region were increased equivalently under these three culture conditions. These results suggest that histone hyperacetylation of the IL-5 gene locus is more sensitive to CD28

FIG. 1. Costimulation with anti-CD28 mAb enhanced the generation of IL-5-and IL-13-producing cells.
A, B, and C, freshly prepared splenic CD4 T cells from B6 mice were cultured with immobilized anti-TCR mAb (H57-597, 3 g/ml) in the presence of costimulation with agonistic anti-CD28 mAb (37.51, 3 g/ml) under Th1-or Th2-skewed conditions for 7 days. The cultured cells were restimulated, and intracellular IL-5/IL-4, IFN␥/IL-4, and IL-5/IL-13 profiles were determined. The percentages of cells present in the each quadrant are shown. D and E, CD4 T cells from normal B6 or STAT6-KO mice were cultured under the conditions described above, and restimulation was done with anti-TCR mAb for 8 h. The amounts of IL-5, IL-13, IL-4, and IFN␥ in the culture supernatant were measured by ELISA. WT, wild type. costimulation than that of the IL-4 locus.
Enhanced Histone Hyperacetylation Induced with CD28 Costimulation Is Observed Only in the IL-5-associated Nucleosomes-Next, we examined the acetylation status of the DNA regions corresponding to IL-5 promoter, IL-5 intron, IL-13 intron, IL-4 promoter, CNS1, IL-4 V A enhancer, RAD50 promoter, and IFN␥ promoter. CNS1 and IL-4 V A enhancer regions were previously described to contain regulatory elements or DNase I hypersensitive sites (28,30). Histone hyperacetylation induced in these regions, except for two controls (RAD50 and IFN␥), occurred in a Th2-specific manner as reported previously (34). As can be seen in Fig. 3A, the levels of acetylation in the region of IL-5 promoter and IL-5 intron were significantly enhanced by the presence of CD28 costimulation. In contrast, those of other regions were all unaffected. The effect of CD28 costimulation on the histone hyperacetylation within the IL-5 gene loci was analyzed more precisely with a series of primer pairs within the IL-5 genes as shown in Fig. 3B. CD28 costimulation significantly enhanced the levels of histone hyperacetylation at all regions within the IL-5 genes tested (Fig.  3C).
Next, we assessed the STAT6 dependence of the CD28 costimulation-induced enhancement of the acetylation (Fig.  3D). Similar to other regions (IL-13 intron, IL-4 promoter, FIG. 2. Induction of histone H3 hyperacetylation of IL-5, IL-4, and RAD50 gene loci in developing Th2 cells with CD28 costimulation. Developing Th1 and Th2 cells cultured with immobilized anti-TCR mAb and CD28 costimulation were prepared 2, 3.5, 5, and 7 days after the stimulation. The acetylation status of histone H3 in the nucleosomes associated with the indicated regions was assessed by a ChIP assay with an anti-acetylated histone H3 antibody and specific primer pairs. Before immunoprecipitation for ChIP assay, aliquots of lysates (ϳ6 ϫ 10 2 cell equivalents) were separated for PCR to determine the relative levels of input DNA. 3-Fold serial dilutions were made with both the input DNA and immunoprecipitated DNA samples before PCR. Two independent experiments were performed, and similar results were obtained. Quantitative representations of the results are shown in the right panels. The intensities of the bands at the highest concentration were measured by densitometry, and relative intensities (anti-acetyl histone precipitates/input DNA ratio) for each primer pair were calculated. CNS1, and IL-4 V A enhancer), histone hyperacetylation of the IL-5 promoter locus was not induced in STAT6-KO T cell cultures even in the presence of CD28 costimulation. Thus, STAT6 is critical for the Th2-specific histone hyperacetylation of the IL-5 gene locus induced by anti-TCR stimulation and CD28 costimulation.
Enhanced Production of IL-5 and Histone Hyperacetylation of the IL-5 Gene Locus Induced by CD28 Costimulation Are Dependent on NF-B Activation-CD28 costimulation is known to induce phosphatidylinositol 3-kinase activation and the subsequent activation of NF-B. We tested the effect of wortmannin, a phosphatidylinositol 3-kinase inhibitor, on the CD28induced enhancement of the generation of IL-5-producing cells (43). The generation of IL-5-producing cells in the culture with CD28 costimulation was decreased in the presence of wortmannin in a dose-dependent manner (Fig. 4A). In contrast, the generation of IL-4-producing cells was not affected at any doses of wortmannin tested. Concurrently, the effect of wortmannin on the production of cytokines was assessed by ELISA (Fig.  4B), and as expected, the enhanced production of IL-5 with CD28 costimulation was sensitive to wortmannin. IL-13 production was also slightly decreased; however, the levels of IL-4 were not changed by the presence of 30 to 300 nM wortmannin.
Next, we assessed the role for NF-B activation in the CD28induced enhancement of the generation of IL-5-producing cells and histone hyperacetylation of the IL-5 gene locus. A mutant form of IB␣ (IB␣M), which inhibits the NF-B activation (44), was inserted in an IRES-GFP retroviral construct, and the The location of the primer pairs used in panel A is as follows (5Ј to 3Ј). IL-5 promoter F, Ϫ521 to Ϫ498; IL-5 promoter R, Ϫ76 to Ϫ101; IL-5 intron F, ϩ857 to ϩ883; IL-5 intron-R, ϩ1214 to ϩ1188. The numbers indicate positions relative to the first nucleotide of the IL-5 exon 1, which is designated as ϩ1. IL-13 intron F, ϩ142/ϩ165; IL-13 intron R, ϩ578 to ϩ555; CNS1 F, ϩ7554 to ϩ7579; CNS1 R, ϩ7767 to ϩ7742. The numbers indicate positions relative to the first nucleotide of the IL-13 exon 1, which is designated as ϩ1. IL-4 promoter F, Ϫ661 to Ϫ637; IL-4 promoter R, Ϫ164 to Ϫ187; IL-4 V A enhancer F, ϩ13234 to ϩ13257; IL-4 V A enhancer R, ϩ13439 to ϩ13416. The numbers indicate positions relative to the first nucleotide of the IL-4 exon 1, which is designated as ϩ1. RAD50 promoter F, Ϫ242 to Ϫ217; RAD50 promoter R, Ϫ43 to Ϫ67. The numbers indicate positions relative to the first nucleotide of the RAD50 exon 1, which is designated as ϩ1. IFN␥ promoter F, Ϫ3372 to Ϫ3349; IFN␥ promoter R, Ϫ3025 to Ϫ3048. The numbers indicate positions relative to the first nucleotide of the IFN␥ exon 1, which is designated as ϩ1. kb, kilobase. C, a ChIP assay with the indicated primer pairs within the IL-5 gene locus was done as in panel A. Three independent experiments were performed with similar results. D, a ChIP assay with STAT6-KO-developing Th2 cells was done as in panel A. Two independent experiments were performed with similar results.

CD28-mediated Enhancement of Hyperacetylation of the IL-5 Gene
vectors were introduced into developing Th2 cells cultured with CD28 costimulation. The expression of the introduced IB␣M was confirmed by immunoblotting with an anti-IB␣ Ab that reacts with both wild type IB␣ and IB␣M (Fig. 4C). Substantial amounts of endogenous IB␣ were detected in the GFP ϩ population of mock pMX-IRES-GFP-infected cells (GFP) and non-infected Th2 cells (Th2). Also, substantial amounts of IB␣M were detected in the IB␣M-infected GFP ϩ population (IB␣M). As previously reported, the upper band, indicated by an arrowhead, is IB␣M (45). The amount of endogenous IB␣ in the IB␣M-infected cells was found to be reduced, probably as a result of the failure of NF-B activation (46). The percentages of IL-5-and IL-4-producing cells in the GFP-positive infected cell population were determined (Fig. 4D). As can be seen, the numbers of IL-5-producing cells were decreased (12.4 Ϯ 23.4 to 6.8 Ϯ 6.1%) by the expression of IB␣M. Interestingly, the percentages of IL-4-producing cells were not significantly affected by IB␣M expression (10.6 Ϯ 12.4 versus 20.5 Ϯ 6.8%). The acetylation status of the IL-5 promoter, IL-4 promoter, and RAD50 promoter regions was assessed in the developing Th2 cells infected with IB␣M vector, and significant down-regulation of hyperacetylation in the IL-5-related nucleosomes was detected (Fig. 4E). Again, the introduction of IB␣M did not inhibit the acetylation levels of the IL-4-and RAD50-related nucleosomes, suggesting that NF-B activation is preferentially involved in the process of hyperacetylation of the IL-5 gene locus.

The Generation of IL-5-producing Cells and Histone Hyperacetylation of the IL-5 Gene Locus Are Highly Dependent on the Expression Levels of GATA3-It is reported that the inhibition of NF-B activity results in reduced GATA3 expression and
Th2 cytokine production in developing but not committed Th2 cells (23). To examine the possible involvement of GATA3 in the CD28-induced enhancement of histone hyperacetylation of the IL-5 gene locus, we assessed the protein expression levels of GATA3 in developing Th2 cells cultured with CD28 costimulation. The GATA3 levels were clearly increased by the presence of CD28 costimulation at the 16-and 32-h time points (Fig. 5A). The increase was abrogated by the presence of wortmannin (Fig. 5B). Furthermore, the transcriptional levels of GATA3 as assessed by semiquantitative RT-PCR were significantly higher in the Th2 cell culture with CD28 costimulation (Fig.  5C). We also examined the transcriptional expression of GATA3 exon 1a and 1b (47). Although the expression of exon 1a transcript was undetectable in these developing Th2 cells, that of exon 1b was moderately enhanced in the presence of CD28 costimulation.
To examine the correlation between GATA3 expression and histone hyperacetylation of the IL-5 gene locus, we introduced GATA3 into CD4 T cells stimulated under Th1-skewed conditions using a retroviral vector (pMX-IRES-EGFP) encoding GATA3 bicistronically with EGFP (pMX-GATA3-IRES-EGFP). The expression of GFP and GATA3 protein in the GATA3infected T cells is depicted in Fig. 5. The expression levels of GATA3 in GFP high (expressing high levels of GATA3, G3) population were ϳ2-fold as compared with those of GFP low (expressing low levels of GATA3, G2) population and equivalent to those in Th2 cells. Next, the levels of IL-5-and IL-4-producing cells were compared between GFP Ϫ (no GATA3 expression, FIG. 6. Long range histone hyperacetylation and intergenic transcripts in the intergenic region of the IL-5 and RAD50 loci in developing Th2 cells with CD28 costimulation. A and B, splenic CD4 T cells were stimulated under the indicated conditions for 7 days, and a ChIP assay was performed. Shown are the PCR product bands for each primer pair (A) and the relative band intensities (B). The results are representative of three independent experiments. The location of GRE-IL-5 is indicated in panel B. kb, kilobase. WT, wild type. C and D, freshly prepared CD4 T cells from B6 and STAT6-KO mice were stimulated under the indicated conditions for 2 days and total RNA was prepared. RNA samples were treated with RNase free DNase I to eliminate any possible genomic DNA contamination, reverse-transcribed (RT ϩ ), and then subjected to PCR with the indicated primer pairs. RT Ϫ represents PCR without reverse transcription. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. The numbers of the primer pairs are the same as those used in panel A. The intensity of bands of the highest concentration was measured, and relative intensities to the ␤-actin bands are shown in panel D. The results are representative of three independent experiments. E and F, the GATA3 introduced cells as in Fig. 5D were sorted (GFP Ϫ , gate G1; GFP low , gate G2; and GFP high , gate G3), and subjected to ChIP assay with indicated primer pairs. The relative intensity (Ac-H3/Input DNA) of each band is shown in panel F. The results are representative of two independent experiments.
To assess the acetylation status of histones in the GATA3introduced developing T cells, GFP Ϫ , GFP low , and GFP high cells prepared as above were purified by cell sorting and subjected to ChIP assay. Histone hyperacetylation of the Th2 cytokine loci (IL-5 promoter, IL-5 intron, IL-13 intron, IL-4 promoter, CNS1, IL-4 V A enhancer) was significantly higher in GATA3-expressing cells (G2 and G3) compared with GATA3 non-expressing cells (G1) (Fig. 5F). The levels of histone hyperacetylation were increased concomitantly with the increase in the expression of GATA3 (compare G2 and G3) in the IL-5 gene locus (IL-5 promoter and IL-5 intron). No such increase was detected in the IL-4-and IL-13-related nucleosomes. These results suggest that the generation of IL-5-producing cells and histone hyperacetylation of the IL-5 gene locus are highly dependent on the expression levels of GATA3.
Long Range Th2-specific Hyperacetylation Detected in the Intergenic Region of the IL-5 and RAD50 Gene Loci Is Enhanced by the Presence of CD28 Costimulation-A series of primer pairs between the IL-5 and RAD50 loci were generated, and the acetylation status of the nucleosomes associated with IL-5 and RAD50 loci was analyzed. The actual band patterns of each ChIP assay (Fig. 6A) and the relative band intensities (Ac-H3/Input DNA) of the 14 selected primer pairs (Fig. 6B) are depicted. A long range Th2-specific hyperacetylation was observed from 400 bp upstream of the RAD50 exon 1 (corresponding to primer 5) to the end of IL-5 exon 4 (primer 14). The acetylation levels of all regions tested were significantly increased in the presence of CD28 costimulation. These results indicate that almost all histones from 400 bp upstream of the RAD50 exon 1 to the end of IL-5 exon 4 (primer 14) are selectively hyperacetylated under Th2-skewed culture conditions and are sensitive to CD28 costimulation. Intergenic Transcription Is Detected throughout the Intergenic Region between the IL-5 and RAD50 Gene Loci-We demonstrated that the intergenic transcription throughout the IL-4 and IL-13 gene loci was accompanied by histone hyperacetylation (34). Thus, we examined the transcription of the intergenic region between the IL-5 and RAD50 gene loci. Interestingly, considerable amounts of transcripts were detected throughout the intergenic region, and the levels were significantly enhanced in the presence of CD28 costimulation (Fig. 6, C and D). In addition, we examined whether the intergenic transcripts were STAT6-dependent or not. STAT6-deficient CD4 T cells were used in paral- These results suggest that intergenic transcripts are induced throughout the intergenic region between the IL-5 and RAD50 gene loci in a Th2-specific and STAT6-dependent manner and are sensitive to CD28 costimulation.
CD28 Costimulation-sensitive Hyperacetylation in the Intergenic Region of the IL-5 and RAD50 Gene Loci Is Dependent on the Levels of GATA3 Expression-Next, we examined the correlation between the levels of GATA3 expression and histone hyperacetylation of the intergenic region. The retrovirus-induced GATA3-expressing cells shown in Fig. 5D were used to compare the acetylation status between GFP Ϫ (no GATA3 expression, G1), GFP low (expressing low levels of GATA3, G2), and GFP high (expressing high levels of GATA3, G3) populations (Fig. 6E). The relative intensity (Ac-H3/input DNA) of each acetylation band of GATA3-introduced cells is shown in Fig.  6F. As expected, the levels of acetylation in the high GATA3expressing cells (G3) were significantly higher than those of low and no GATA3-expressing cells (G2 and G1, respectively), suggesting that histone hyperacetylation of the intergenic region requires a high level expression of GATA3.

DISCUSSION
In this report we demonstrated that CD28 costimulation controls Th2-specific histone hyperacetylation of the IL-5 gene locus. CD28-mediated activation of NF-B and the resulting enhancement of GATA3 induction appeared to be a mechanism by which histone hyperacetylation of the IL-5 gene locus was efficiently induced. This regulation was IL-5 gene-specific because the effect of CD28 costimulation was not observed in the acetylation of the IL-13 or IL-4 gene loci. A long range CD28sensitive histone hyperacetylation with transcripts was detected in the IL-5 and intergenic region between the IL-5 and RAD50 gene.
The generation of IL-5-and IL-13-producing cells and the production of these cytokines were enhanced by CD28 costimulation of the differentiation culture (Fig. 1). A similar conclusion was drawn from the experiments with wortmannin (Fig. 4,  A and B). As for histone hyperacetylation, however, CD28 costimulation affected only the IL-5 gene locus ( Fig. 2 and 3). The transcription of IL-5 and IL-13 is known to be highly dependent on GATA3 as compared with that of IL-4 (11,48,49). An efficient transcription of IL-5 or IL-13 may require the enhanced levels of GATA3 that can be achieved by the presence of CD28-costimulation. Thus, it is possible that CD28 costimulation enhanced both histone hyperacetylation and transcription at the IL-5 gene locus but enhanced only transcription at the IL-13 gene. However, it would be unlikely that the enhancement of IL-5 and IL-13 production is mainly due to the effect on transcription, because we did not include anti-CD28 costimulation when the differentiated Th2 cells were restimulated. In fact, the production of IL-4 and IL-5 was only marginally increased when differentiated Th2 cells were restimulated with anti-TCRϩanti-CD28. 2 This is consistent with the results reported previously (23).
NF-B was reported to interact with histone acetyltransferases such as CREB-binding protein/p300 (50 -52). In addition, NF-B binding influenced the recruitment of SWI/SNF-type chromatin remodeling complexes in the granulocyte-macrophage colony-stimulating factor promoter in T cells (53). Thus, it is conceivable that CD28-induced NF-B activation is involved directly in the acetylation of the IL-5 gene locus at the chromatin level. However, there is no NF-B binding motif in the intergenic region of the IL-5 and RAD50 gene loci except for one in the promoter region of the IL-5 gene. Thus, it is most likely that the enhanced histone hyperacetylation of the IL-5 gene locus induced by the presence of CD28 costimulation is due to the enhanced expression of GATA3. NF-B induces a wide variety of genes, such as cytokines (e.g. tumor necrosis factor-␣ and granulocytemacrophage colony-stimulating factor), chemokines (e.g. MCP-1 (monocyte chemoattractant protein)), RANTES (regulated on activation normal T cell expressed and secreted), and eotaxin), and adhesion molecules (e.g. ICAM (intercellular adhesion molecule 1) and VCAM (vascular cell adhesion molecule 1) (54,55). Thus, it is also possible that other genes regulated by NF-B activation play critical roles in the histone hyperacetylation of the IL-5 gene locus; however, further investigation is required for addressing this issue.
We detected a long range histone hyperacetylation accompanying intergenic transcripts throughout the intergenic region of the IL-5 and RAD50 gene loci (Fig. 6). This is reminiscent of the GATA3-dependent hyperacetylation of the IL-13 and IL-4 gene loci (34,42), suggesting that a similar molecular mechanism governs the acetylation events of both IL-13/IL-4 and IL-5 genes. The difference was the sensitivity to CD28 costimulation and the dependence on the levels of GATA3. Although the reason for the difference is not clear at this time, the nature of putative GATA response elements responsible for the IL-5 gene acetylation could be distinct from that of conserved GATA3 response element (34). There is 60% homology in the DNA sequence around the upstream region of human RAD50 gene compared with mouse, but we did not identify any conserved GATA binding motifs. However, there are several GATA binding motifs present in both mouse and human, suggesting a possible targeting of GATA3 to this region.
Hyperacetylation of the histone H3 (K9/14) and H4 (K5/8/12/ 16) is associated with transcriptionally active chromatin (33). However, acetylation of the histone H3-K9/14 does not always correlate with histone H4 acetylation (56). Furthermore, methylation of histone H3-K4 appears to be correlated with active chromatin (57). In the study we focused on the acetylation status of histone H3-K9/14. Thus, further analysis of histone H4 and histone H3-K4 methylation will be required to provide a more detailed view of the chromatin remodeling of the IL-5 gene locus.
In conclusion, we have demonstrated a possible molecular mechanism that controls histone hyperacetylation of the IL-5 gene locus. Characteristic features of chromatin remodeling of the IL-5 gene locus as compared with those of IL-13 and IL-4 were revealed to be the differential involvement of CD28 costimulation and sensitivity to the levels of GATA3 protein. This study is the first to provide evidence that CD28 costimulation controls chromatin remodeling during Th2 cell differentiation.