IFN-α suppresses GATA3 transcription from a distal exon and promotes H3K27 tri-methylation of the CNS-1 enhancer in human Th2 cells

Abstract

CD4⁺ T helper type 2 (Th2) development is regulated by the zinc finger transcription factor GATA3. Once induced by acute priming signals, such as IL-4, GATA3 poises the Th2 cytokine locus for rapid activation and establishes a positive feedback loop that maintains elevated GATA3 expression. Type I interferon (IFN-α/β) inhibits Th2 cells by blocking the expression of GATA3 during Th2 development and in fully committed Th2 cells. In this study, we have uncovered a unique mechanism by which IFN-α/β signaling represses the GATA3 gene in human Th2 cells. IFN-α/β suppressed expression of GATA3 mRNA that was transcribed from an alternative distal upstream exon (1A). This suppression was not mediated through DNA methylation, but rather by histone modifications localized to a conserved non-coding sequence (CNS-1) upstream of exon 1A. IFN-α/β treatment lead to a closed conformation of CNS-1 as assessed by DNase I hypersensitivity along with enhanced accumulation of H3K27me3 mark at this CNS region, which correlated with increased density of total nucleosomes at this putative enhancer. Consequently, accessibility of CNS-1 to GATA3 DNA binding activity was reduced in response to IFN-α/β signaling, even in the presence of IL-4. Thus, IFN-α/β disrupts the GATA3 autoactivation loop and promotes epigenetic silencing of a Th2-specific regulatory region within the GATA3 gene.

Introduction

GATA3 is a critical transcriptional regulator involved in a variety of cellular differentiation pathways. In the immune system, GATA3 is required for hematopoiesis, thymic development, and peripheral T cell effector functions (1). GATA3 is a critical regulator of the Th2 phenotype, and its elevated expression in T cells is required for both Th2 development and for maintaining the stability of Th2 memory cells (2–4). Although GATA3 is expressed at basal levels in naive T cells, modest increases in GATA3 protein levels can promote Th2 commitment even under a variety of conditions that drive other phenotypes (5). Moreover, early studies by Murphy and colleagues demonstrated that ectopic expression of GATA3 via retroviral transduction led to the induction of GATA3 mRNA encoded by the endogenous gene (3). These data suggested a mechanism whereby GATA3 autoactivation could not only drive Th2 development but also maintain the Th2 phenotype in the absence of further acute developmental signals such as IL-4 (6). Formal proof for the requirement of GATA3 in maintaining the Th2 program was demonstrated by deleting GATA3 in fully committed mouse and human Th2 cells (7, 8). Thus, GATA3 plays a dominant role in maintaining the stability of Th2 cells, and any pathway that suppresses its expression would be predicted to inhibit Th2 functions.

Recently, we (9) and others (10) demonstrated that, unlike IL-12 or other innate cytokines, type I interferon (IFN-α/β) blocked IL-4-mediated Th2 development in human T cells and destabilized the Th2 phenotype by suppressing IL-4, IL-5 and IL-13 secretion. This effect, however, was not observed in murine T cells (9, 11). Further, we found that the inhibition was mediated by suppressing GATA3 expression during Th2 development and in committed Th2 cells. In this study, we found that IFN-α/β suppressed GATA3 by selectively targeting the expression of the GATA3 gene at an alternative upstream exon (1A) utilized in response to IL-4 during Th2 commitment. The repression of exon 1A correlated with a condensed chromatin conformation of a conserved non-coding sequence (CNS-1) region located 5 kb upstream of the alternative exon. Thus, epigenetic silencing of a putative enhancer of the Th2-specific GATA-3 exon 1A promoter is a potential target for the induction of tolerance in atopic Th2 cells.

Materials and Methods

Human Subjects

Peripheral blood was obtained from healthy adults by venipuncture. Informed consent was obtained from each donor in accordance with guidelines established by the IRB at UT Southwestern Medical Center.

Cell Culture and Reagents

Human naïve T cells (CD4+/CD45RA+) were purified (≥90%) from buffy coats either by flow cytometric sorting or by magnetic bead separation. Cells were activated with plate-bound anti-CD3 (OKT3, 3 μg/ml), anti-CD28 (3 μg/ml), and IL-2 (50 U/ml) in complete IMDM supplemented with 10% FBS under the following polarizing conditions: Neutralized (anti-IL-4 (2 μg/ml), anti-IL-12 (5 μg/ml), anti-IFN-γ (5 μg/ml), and anti-IFNAR2 (2 μg/ml)), IL-4 (IL-4 (20 ng/ml), anti-IL-12 (5 μg/ml), anti-IFN-γ (5 μg/ml), and anti-IFNAR2 (2 μg/ml)), IFN-α (IFN-α(A) (1000 U/ml), anti-IL-4 (2 μg/ml), anti-IL-12 (5 μg/ml), anti-IFN-γ (5 μg/ml)), and IL-4 + IFN-α (IL-4 (20 ng/ml), IFN-α(A) (1000 U/ml), anti-IL-12 (5 μg/ml), and anti-IFN-γ (5 μg/ml)), In some experiments, the following inhibitors were used: MG132 (50 μM), and 5-Azacytidine (1 μM). Cells were cultured for 3, 5, or 7 days prior to being used for analysis.

Flow Cytometry

Intracellular cytokine staining was performed as previously described (12). Cell proliferation was assessed using CFSE at day 5. For live cell sorting of CFSE-labeled cells for mRNA analysis on day 5, CFSE-labeled cells were washed and sorted on a MoFlo or a FACSAria cell sorter. RNA was harvested and used for qPCR analysis.

Fluorescence Microscopy

Cells were adhered to coverslips and fixed with 4% PFA, then stained with anti-hCD4-PE and anti-hGATA3-Alexa647. Coverslips were mounted onto slides with Prolong Gold + DAPI (Invitrogen). Fluorescence microscopy was performed on a Deltavision deconvolution microscope and images were prepared using ImageJ software (NIH).

Quantitative PCR

qPCR was performed as described (13). Human PPIA was used as a reference gene. Relative changes in mRNA expression were calculated by the 2^−ΔΔCT method. All treatments were referenced to the neutralized control. Primers can be found in Supplemental Table 1.

Electrophoretic mobility shift assay (EMSA)

Clarified nuclear lysate was incubated with 3′-biotin-labeled dsDNA probes and resolved on a 4.5% non-denaturing polyacrylamide gel, then transferred to nylon membranes. Complexes were detected with streptavidin-HRP with chemilluminescence. Oligo sequences can be found in Supplemental Table 1.

Chromatin Immunoprecipitation (ChIP)

Cells were fixed with 1% formaldehyde, and nuclear DNA was sheared by sonication. Pre-cleared lysates were incubated overnight with 1–4 ug of antibody, followed by incubation with protein A/G beads. Crosslinks were reversed from eluted protein/DNA complexes, and remaining protein was degraded with Proteinase K. Purified DNA was used for qPCR analysis, and primer sequences can be found in SI Table 1. ChIP efficiency was calculated using the following formula: $2^{(\text{Input}\text{Ct}-\text{ChiP}\text{Ct})}\times \text{dilution}\text{factor}$. All primer and probe sequences are listed in Supplemental Table 1.

DNase I hypersensitivity analysis

In vitro primed human CD4⁺ T cells were harvested on day 5 of culture and washed in cold PBS. Nuclei were permeabilized with 0.3% NP-40 in in Buffer A (15 mM Tris-HCl pH 8.0, 15 mM NaCl, 60 mM KCl, 1mM EDTA, 0.5 mM EGTA, 0.5mM spermidine, and 0.15 mM spermine) for 10 min on ice, rinsed once, then incubated with DNase I (0 – 120 U/ml, Worthington) at 37°C for 5 min. The reactions were stopped with the addition of Stop Buffer (50 mM Tris-HCl pH 8.0, 100 mM NaCl, 1% SDS, 100 mM EDTA) supplemented with RNase A. Nuclei were digested with proteinase K at 55°C for 18 h followed by DNA extraction. Semi-quantitative PCR analysis of incrementally tiled segments spanning across GATA3 CNS-1 through exon 1A were performed, and the primer sequences for each segment are listed in Supplemental Table 1. DNA band intensities were quantified by Image J software, and relative DNase hypersensitivity was calculated by comparing the slope of DNA decay between cytokine treated and neutralized samples. qPCR of specific regions was performed as described above with the primers listed in Supplemental Table 1. The slope of the line formed by the reduction in Ct value as a function of DNase I concentration was calculated, and relative DNase I hypersensitivity was calculated as a ratio of slopes. Data for each sample were referenced to the Neutralized control.

Statistical analysis

All graphs are shown as mean +/− SEM. Statistical analysis was performed by one-way or two-way ANOVA with Graphpad Prism software. Values of p < 0.05 were considered significant.

Results

Inhibition of GATA-3 expression by IFN-α/β

The GATA3 transcription factor is expressed during all stages of thymic development and remains expressed constitutively in resting peripheral naïve CD4⁺ T cells. Cells are prevented from committing to the Th2 due the stoichiometric expression of the ROG-1/FOG-1 repressors (14, 15), which prevent GATA3 from driving their differentiation. However, in response to IL-4, GATA3 levels are elevated sufficiently to overcome this repression. We recently demonstrated that IFN-α/β blocked the induction of GATA3 in response to IL-4 (9). This effect was confirmed in human CD4⁺/CD45RA⁺ T cells differentiated in response to IL-4 in the absence or presence of IFN-α (Fig. 1A and B). Here, IL-4 increased GATA3 protein levels less that two-fold, which was completely suppressed by treatment with IFN-α (Fig. 1A). Further, this suppression was seen at the RNA level, where as little as 100 U/ml of IFN-α significantly reduced IL-4-driven GATA3 induction (Fig. 1B).

Fig. 1.

IFN-α/β suppresses IL-4-driven induction of GATA3 expression as a function of cell division. (A) Purified human CD4⁺/CD45RA⁺ cells were activated with plate-bound anti-CD3/anti-CD28 for 6 days under the indicated cytokine conditions. Cell lysates were assessed by Western blotting for GATA3 and reprobed for β-tubulin, Bands were quantified by densitometry, and ratios of GATA3/β-tubulin are listed beneath each lane. (B) Purified human CD4⁺/CD45RA⁺ cells were activated with IL-4 and increasing concentrations of IFN-α (10–1000 U/ml) for 3 days. GATA3 mRNA was quantified by qPCR, and data are expressed relative to the neutralized control. (C) Purified human CD4⁺/CD45RA⁺ cells were labeled with CFSE and activated as described above. Cells were restimulated with PMA/ionomycin in the presence of monensin and stained for intracellular GATA3 and IL-4. Data are gated on live cells and the polygon gate indicates a threshold of GATA3 expression based on basal levels of GATA3 staining in the neutralized condition. (D) The mean fluorescence intensity (MFI) was calculated for each CFSE division peak. (E) Data from (C) were gated on CFSE division 4 and assessed for expression of GATA3 and IL-4.

During the early phases of innate priming, T cells divide rapidly in response to antigen receptor and costimulatory signals. Innate cytokines drive their differentiation into effector cells, which occurs progressively at each cell division (4, 16–19). During Th2 development, the ability to secrete IL-4 increases incrementally in daughter cells as they divide in the presence of IL-4. As IFN-α inhibits the Th2 differentiation process by suppressing GATA3 (9), we wished to determine whether IL-4-driven GATA3 expression was inhibited by IFN-α by slowing the progression of cell division or by direct repression of GATA3 in daughter cells. To test this, human CD4⁺/CD45RA⁺ cells were activated with anti-CD3/CD28 in the presence or IL-4, IFN-α, or IL-4 + IFN-α, and GATA3 protein levels were quantified as a function of CFSE dilution (Fig. 1C). At day 5 of activation, a similar proportion of divided cells at each division were observed, regardless of the innate cytokine priming condition imposed at the beginning of the culture. GATA3 expression was enhanced incrementally at each cell division (Fig. 1D), and IFN-α completely inhibited the induction of GATA3 protein by IL-4. The inhibition of GATA3 by IFN-α correlated with a reduction in the percentage of cells that secreted IL-4 upon restimulation (Fig. 1E). Thus, IFN-α inhibited the induction of GATA3 by IL-4 without significantly altering cellular expansion in response to TCR stimulation.

Since IL-4 signaling establishes a positive feedback loop of GATA3 autoactivation (3, 4), there are several steps at which IFN-α could block this pathway. For example, IFN-α could suppress the initial induction of GATA3 by down-regulating the IL-4 receptor or by interfering with STAT6 activation (20, 21). However, our recent studies ruled out this possibility (9), and our observations that IFN-α can repress GATA3 expression in fully committed Th2 cells makes this scenario unlikely. Further downstream, IFN-α could disrupt GATA3 autoactivation through a variety of post-translational mechanisms. First, GATA3 nuclear localization was examined by confocal microscopy in cells differentiated in vitro with IL-4 or IFN-α by confocal microscopy (Fig. 2A). Although IFN-α decreased the total staining intensity of GATA3, all of the detectable GATA3 was co-localized with DAPI staining under all cytokine conditions, suggesting that IFN-α did not block GATA3 autoactivation by excluding GATA3 from the nucleus. Second, IFN-α could lead to proteosomal degradation of GATA3 protein. However, treatment of cells with the general proteosome inhibitor, MG132, failed to reverse the suppressive effects of IFN-α (Fig. 2B). Other post-translational modifications could interrupt the ability of GATA3 to bind DNA and regulate transcription. We tested the ability of IFN-α to alter GATA3 DNA binding activity by both EMSA and ChIP. First, IL-4 significantly enhanced binding of GATA3 to a consensus GATA3 target sequence, which was completely inhibited by IFN-α (Fig. 2C and D). Of note, the enhancement of GATA3 DNA binding activity by IL-4 was greater in magnitude (> 3 fold) than what can be accounted for by a modest (< 1.5 fold) induction of total GATA3 protein, perhaps suggesting additional mechanisms of regulation in addition to simple increases in protein content. Further, the inhibition of GATA3 binding activity by IFN-α was paralleled by a reduction of GATA3 bound to the canonical GATA3 site found within the first intron of the IL-4 gene (Fig. 2E). Collectively, these data rule out the possibility that IFN-α blocks GATA3 expression by either preventing nuclear import or by proteosomal degradation. However, the reduction in GATA3 DNA binding activity could be due to overall reductions in GATA3 protein content, which may be regulated transcriptionally rather than at the post-translational level.

Fig. 2.

IFN-α/β inhibits GATA3 DNA binding activity. Purified human CD4⁺/CD45RA⁺ cells were activated for 5 days with plate-bound anti-CD3/anti-CD28 with the indicated cytokine conditions. (A) Cells were stained with DAPI (blue), anti-CD4 (green) and anti-GATA3 (red) antibodies and visualized by confocal microscopy. (B) Cells were incubated in the absence or presence of MG132 and intracellular GATA3 was quantified by flow cytometric staining. (C) GATA3 DNA binding activity from nuclear extracts was measured by EMSA with a labeled consensus GATA3 DNA probe from the IL-4 CIRE. Specificity controls included cold competitor (CC) and anti-GATA3 super shift (SS). The EαY box probe served as a loading control. (D) DNA binding complexes in (C) were quantified by densitometric scanning. (E) GATA3 DNA binding at the IL-4 CIRE was assessed by ChIP, and data are expressed relative to the neutralized control. *, p ≤ 0.05.

IFN-α/β-mediated transcriptional repression of an alternative GATA-3 distal exon

The GATA3 gene contains two independently regulated first exons (22, 23), denoted 1A and 1B (Fig. 3A). These exons are separated by approximately 10 kb, and when transcribed, each first exon is spliced to exon 2, which contains the initiator codon. Splicing of exons 1A with 1B has not been detected in any cell type, which may indicate that the two exons operate as distinct transcriptional units. In support of this, previous studies identified exon 1A to be selectively induced by IL-4 in murine peripheral CD4⁺ T cells, leading to Th2 development (22). We tested whether induction of GATA3 exon 1A was selectively induced by IL-4 in human T cells. Indeed, IL-4 increased total GATA3 mRNA content as assessed by qPCR with primers that spanned across exons 5 and 6 (Fig. 3B, upper panel). As previously demonstrated, IFN-α inhibited total GATA3 mRNA, which correlated with the decrease in GATA3 protein content (Fig. 3B, upper panel). Further, IL-4 selectively induced exon 1A, but not 1B, and this induction was also blocked by IFN-α (Fig. 3B, middle and bottom panels). In contrast, exon 1B was not affected by any cytokine condition and may be the main exon that contributes to constitutive low expression of GATA3 in naïve T cells.

Fig. 3.

Progressive expression of GATA3 exon 1A and inhibition by IFN-α/β during cell division. (A) GATA3 mRNA transcription can proceed from two alternative first exons, 1A and 1B, each controlled by separate promoters. Arrows indicate the direction of primers to distinguish 1A from 1B by qPCR. The CNS-1 site is positioned approximately 5 kb upstream of exon 1A. (B) Purified human CD4⁺/CD45RA⁺ cells were activated for 3 days with plate-bound anti-CD3/anti-CD28 with the indicated cytokine conditions. GATA3 mRNA transcripts were quantified by qPCR with primers that measured total GATA3 spanning exons 5–6 (upper panel), and with primers that distinguished exons 1A (middle panel) from 1B (lower panel). (C) CFSE labeled human CD4⁺/CD45RA⁺ cells were activated as above for 5 days, and individual cell divisions were sorted based on the gates shown in the upper panel. GATA3 mRNA transcripts derived from exon 1A (middle panel) and exon 1B (lower panel) were quantified by qPCR and expressed relative to the neutralized Div 0 population. (D) Purified human CD4⁺/CD45RA⁺ cells were activated with plate-bound anti-CD3/anti-CD28 for 7 days under the indicated cytokine conditions for primary activation (1°). Cells were then washed and restimulated for an additional 7 days with the cytokine conditions indicated for secondary activation (2°). GATA3 exon 1A and 1B transcripts were quantified by qPCR.

We further assessed GATA3 mRNA expression in dividing populations of cells undergoing differentiation in response to IL-4 or IFN-α (19, 24). Cells were labeled with CFSE, activated with anti-CD3/CD28 for 5 days, and then purified by sorting based on CFSE dilution as indicated by the gates in Fig. 3C. In parallel with the expression of GATA3 protein (Fig. 1C and D), we found that GATA3 exon 1A was selectively induced by IL-4 at each incremental cell division. Exon 1B was not altered either by cell division or by cytokines. In addition, we analyzed exon 1A and 1B expression in committed Th2 cells. Once induced by IL-4, GATA3 autoactivation stabilizes Th2 cells by uncoupling their phenotype from the initial signals that drove their development. Further, our previous studies demonstrated that IFN-α could suppress GATA3 expression in fully committed Th2 cells thus disrupting the overall Th2 program (9). As shown in Fig. 3D, IL-4 induced the expression of GATA3 exon 1A which remained elevated in subsequent rounds of stimulation regardless if IL-4 was neutralized in the second week. However, in agreement with our previous findings, IFN-α suppressed GATA3 exon 1A expression in committed Th2 cells down to levels observed in cells that were activated under neutralizing conditions. In summary, we found that the induction of GATA3 mRNA by IL-4 was regulated exclusively at exon 1A, increased as a function of cell division, and was stabilized in differentiated Th2 cells. Importantly, this regulation was blocked by IFN-α, which suppressed GATA3 exon 1A expression even in committed Th2 cells.

Epigenetic modification of GATA-3 CNS-1 in human Th2 cells

Located upstream of exon 1A are several conserved non-coding sequences that have been identified previously by VISTA analysis (23). The most proximal CNS-1 region is positioned 5 kb upstream of exon 1A (Fig. 4A) and is of particular interest as it contains multiple GATA3 consensus binding sites, which may be targets for GATA3 autoactivation. We wished to determine how this genomic region was being regulated both positively by IL-4 and negatively by IFN-α/β. We assessed DNase hypersensitivity as well as various primary chromatin modifications that directly impact transcription including DNA methylation and histone acetylation/methylation. First, given that local increases in DNA methylation patterns often correspond with reductions in transcriptional activity, we proposed that CNS-1 could be a target of IFN-α-mediated recruitment of DNA methyltransferases (25). However, the general DNA methylation inhibitor 5-AzaC did not prevent IFN-α from markedly suppressing GATA-3 exon 1A expression (Supplemental Fig. 1), suggesting that IFN-α was able to suppress GATA-3 mRNA expression independently of the DNA methylation status of the cells or the GATA-3 locus specifically.

Fig. 4.

IFN-α/β signaling selectively decreases DNase I relative hypersensitivity at the GATA3 CNS-1 region and the exon 1A transcriptional start site. (A) Purified human CD4⁺/CD45RA⁺ cells were activated with plate-bound anti-CD3/anti-CD28 for 5 days under the indicated cytokine conditions. Cells were permeabilized and incubated with increasing concentrations of DNase I. DNA was purified, and semiquantitative PCR analyses were performed with primers spanning 10 overlapping tiled intervals across the CNS-1/exon 1A region indicated by the gray bars in the diagram (Supplemental Table 1). Amplicons were resolved by gel electrophoresis (Supplemental Fig. 2), and band intensities were quantified by densitometry using ImageJ software. The slope of the line formed by the reduction in amplicon as a function of DNase I concentration was calculated, and relative DNase I hypersensitivity was calculated as a ratio of slopes of Neutralized versus cytokine treatment. (B) Quantitative DNase I hypersensitivity analysis of CNS-1 region was performed as described above from 4 healthy adult donors. Samples 4a and 4b are separate experiments performed on cells from the same donor. Relative hypersensitivity is referenced to the Neutralized control of each donor/experiment. (C) Averaged data from (B) are expressed relative to the Neutralized control. *, p ≤ 0.05.

Local chromatin compaction at both promoters and enhancers can significantly alter transcription rates. Such regions can be qualitatively compared by their accessibility to digestion with DNase I. As a first approach, we assessed relative DNase I hypersensitivity across the entire CNS-1/Exon 1A region of the GATA3 gene as a function of cytokine treatment. Primary naïve human CD4⁺ T cells were differentiated in vitro in the presence of IL-4 or IL-4 + IFN-α and compared to cells differentiated under neutralizing conditions. DNase I hypersensitivity analysis was performed on these cells by interrogating tiled intervals across the CNS-1/Exon 1A region by semi-quantitative PCR analysis (Fig. 4A and Supplemental Fig. 2). We identified two regions in which relative DNase I hypersensitivity was reduced in the presence of IFN-α. The first region mapped across tiling interval 1, which spanned the entirety of CNS-1. The second region spanned intervals 9–10, corresponding to the transcriptional start site and first intron of exon 1A (Fig. 1A). We did not observe any cytokine-mediated changes in DNase I hypersensitivity within the 5′ UTR of exon 1A spanning from the putative promoter (interval 8) through the most distal region adjacent to CNS-1 (interval 2). As CNS-1 has been identified as a critical regulatory element of GATA3 expression in murine Th2 cells, we confirmed the DNase I hypersensitivity of CNS-1 in T cells isolated from multiple donors by a quantitative PCR assay. IL-4 enhanced the relative DNase I hypersensitivity of CNS-1 in T cells from 4 separate donors, whereas IFN-α suppressed this activity to levels comparable to the neutralized control (Fig. 4B and C). Thus, IFN-α promoted a closed chromatin configuration of CNS-1 and potentially at the exon 1A transcriptional start site. These data suggest a role for IFN-α/β signaling in suppressing the transcriptional accessibility of this region during human Th2 commitment.

Changes in specific histone modifications such as acetylation and methylation can often distinguish or predict regions of enhanced or suppressed transcriptional activity. We quantified various histone modifications by ChIP at the GATA-3 CNS-1 region as well as local segments spanning the putative promoter region of exon 1A (Fig. 5A). For these experiments, ChIP of total histone H3 along with H3K27me3, H3K9me2, and H4Ac were performed in CD4⁺ T cells isolated from 4 or 5 separate donors, depending upon the modification. The relative ChIP efficiencies for each donor were calculated and expressed as percent of the neutralized control. For H3K9me2 (Fig. 5A), none of the cytokine conditions led to a significant alteration in the density of these marks at any of the sites interrogated. The H4Ac modification was significantly increased at proximal promoter sites by IL-4 regardless if IFN-α was present during the priming, whereas this modification was not significantly altered at CNS-1 (Fig. 5A, lower panel). However, the H4Ac modification pattern at the exon 1A promoter cannot explain the dominant effect of IFN-α in suppressing GATA3 expression in the presence of IL-4. In assessing repressive marks, we found that H3K27me3 was not significantly altered by cytokine activation at the exon 1A promoter. In contrast, we found that the H3K27me3 modification was significantly enriched at the CNS-1 site in response to IFN-α in both the absence and presence of IL-4, which correlates with the inhibition of exon 1A expression by IFN-α. The increased H3K27me3 modification was reflected by a significant increase in the total density of H3 in response to IFN-α at CNS-1, but not the exon 1A promoter (Fig. 5A, upper panels). Thus, the repressive marks at CNS-1 correlate well with the reduction of DNase I hypersensitivity at this region, implicating IFN-α/β signaling in blocking this potential enhancer element.

Fig. 5.

Epigenetic silencing of GATA3 CNS-1 in response to IFN-α/β. (A) ChIP was performed with CD4⁺/CD45RA⁺ T cells activated for 5 days as described above, and the data are expressed as percent control of the ChIP efficiencies of each of the Neutralized control samples. The relative density was assessed for total H3, H3K27me3, H3K9me2, and H4Ac within CNS-1 and three adjacent sites across ~1 kb of sequence upstream of GATA3 exon 1A. The four PCR primer sites are indicated by the gray bars beneath the diagram and are denoted I–IV. Data are averages from 4 or 5 independent healthy adult control blood donors (*p < 0.05). (B) GATA3 DNA binding to the GATA3 CNS-1 region was assessed by ChIP in purified CD4⁺/CD45RA⁺ T cells activated for 5 days with the indicated cytokine conditions. The PCR primers spanned two adjacent regions (CNS-1a and 1b) within the 650 bp CNS-1 site and are indicated by the gray bar in the upper diagram along with putative GATA3 consensus binding sites (flags). Experiments from two separate healthy adult blood donors are shown.

The CNS-1 region contains at least 8 conserved GATA3 consensus binding sites (Fig. 5B), which are potential targets for GATA3 autoactivation. Since IFN-α/β led to a closed confirmation of this region (shown above) we wished to determine whether this site was bound by GATA3, and if so, whether this binding was blocked by IFN-α/β treatment. To test this, GATA3 ChIP was performed at CNS-1 in human CD4⁺ T cells differentiated in the presence of IL-4 or IFN-α. IL-4 significantly increased GATA3 binding to both segments of the CNS-1 region spanning 650 bp, both of which contained consensus GATA3 binding sequences (Fig. 5B). Further, IFN-α inhibited IL-4-driven binding of GATA3 to CNS-1 in T cells from both donors. We also assessed GATA-3 binding at the exon 1A proximal promoter, but were unable to detect any significant GATA-3 binding activity under any condition. Thus, the CNS-1 site may contribute to GATA3 exon 1A regulation by acting as an enhancer for GATA3 autoactivation. Further, IFN-α/β represses this activity, which correlates with epigenetic silencing of CNS-1 even in the presence of IL-4.

In summary, we found that DNA binding by GATA3 at the CNS-1 region increased in response to IL-4 but was inhibited by IFN-α. This activity coincided with enrichment of the suppressive H3K27me3 mark selectively at the CNS-1 site, but not the exon 1A promoter. The reduction in DNAase I hypersensitivity, the increase in the H3K27me3 mark, and consequently the increase in total H3 content selectively at CNS-1 strongly suggests that this area becomes more compact and inaccessible in response to IFN-α signaling. Collectively, these data highlight the selective repressive activity of IFN-α on GATA3 exon 1A transcription in human CD4⁺ T cells.

Discussion

In mouse CD4⁺ T cells, various external stimuli, such as IL-4 and Notch, can promote elevated expression of GATA3, which is selectively encoded from the alternative exon 1A within the GATA3 locus (22, 23, 26). The present study is the first to demonstrate that IL-4 preferentially induces GATA3 exon 1A expression during human Th2 development. Furthermore, we show that the exon 1A transcript, but not the exon 1B transcript, remains elevated in Th2 cells in the absence of further IL-4 signaling, suggesting that GATA3 feedback in human Th2 cells preferentially maintains expression of the exon 1A transcript. Moreover, we identified the CNS-1 region as a potential GATA3 autoregulatory enhancer element that was bound by GATA3 in response to IL-4. Finally, we uncovered an IFN-α/β-dependent mechanism that suppressed Th2 development and stability by disrupting the GATA3 autoactivation loop. We systematically ruled out defects in nuclear localization or proteosome-mediated degradation while finding that overall GATA3 DNA binding activity was reduced. However, a reduction in GATA3 DNA binding activity could be due to reduced levels of GATA3 present in the nucleus of IFN-α treated cells. Alternatively, IFN-α signaling could alter the function of the GATA3 protein by inhibiting the ability of GATA3 to regulate its own expression. However, this possibility is somewhat unlikely since IFN-α also blocks the induction of GATA3 expression by IL-4 prior to the establishment of the autoactivation loop. Finally, IFN-α could inhibit transcriptional activation of the GATA3 gene, which is supported by our data demonstrating a selective block in GATA3 exon 1A expression both in response to IL-4 and in fully committed Th2 cells.

IFN-α/β is a potent inducer of hundreds of interferon-sensitive genes that regulate the antiviral response, but very few genes are actually suppressed by IFN-α/β signaling. Of the few select genes that are repressed by IFN-α/β, some are involved in cell cycle regulation, such as cyclins (27, 28). As such, we have previously shown that IFN-α slows the progression of cell division in human CD8⁺ T cells, thus preventing some cells from terminally differentiating into effectors (19, 24). However, in this study, we found that IFN-α did not have this effect on cells in the CD4⁺ T cell compartment. Rather, IFN-α suppressed GATA3 expression without significantly altering TCR-mediated proliferation, thus blocking IL-4-driven terminal differentiation of Th2 cells. Although cell division is not a “clock” that strictly controls cytokine production (18), S phase offers the best opportunity to modify local chromatin architecture and alter its accessibility. In this regard, we observed a marked loss in DNase I hypersensitivity, which correlated with a significant increase in the total density of histone H3 specifically at CNS-1 in response to IFN-α. There are three forms of histone H3 (H3.1, H3.2, and H3.3), which are encoded by distinct genes. Although our study design did not distinguish between these H3 variants, recent reports have suggested that H3.3 in particular can be deposited at nucleosome-depleted gaps in chromatin (29). H3.3 is usually associated with areas of active transcription, but has also been found in repressed and poised regions (30, 31). Furthermore, changes in nucleosome density have been reported as a mechanism that controls epigenetic modifications of the histones (32), suggesting that nucleosome density could be regulated through a mechanism distinct from direct modification of the histones. In most cases, IFN-α/β signaling generally promotes transcription rather than repressing it, which is accounted for by specific deposition of histones H3.3 at interferon-sensitive genes (ISGs) (33). However, we find that IFN-α/β selectively repressed expression of GATA3 in a manner that may be distinct from the modes of transcriptional regulation of ISGs.

Based on our current findings, we propose a unique model of GATA3 regulation that accounts for both the induction by IL-4 and repression by IFN-α/β (Fig. 6). The region encompassing CNS-1 displays different epigenetic patterns in response to cytokine activation than the regions more proximal to the exon 1A promoter. The permissive mark H4Ac is increased by IL-4 but reduced by IFN-α in the absence of IL-4. Although IL-4 increases H4Ac even in the presence of IFN-α at the exon 1A promoter, this does not occur at CNS-1. Furthermore, IFN-α reduced H4Ac levels below baseline near exon 1A but not at CNS-1. This is in stark contrast to the repressive H3K27 mark in which IFN-α treatment, even in the presence of IL-4, increased H3K27me3 at CNS-1 but induced no significant changes at the exon 1A promoter. That IL-4 signaling does not result in significant changes in permissive marks at CNS-1 is even more surprising considering that this region is bound by GATA3. GATA3 can complex with trithorax group proteins that normally increase chromatin accessibility (34, 35). As such, GATA3 binding to CNS-1 would be predicted to increase permissive marks on the histones and reduce nucleosomal density. However, as IL-4 treatment marginally increased DNase I hypersensitivity, it did not reduce H3 density. Thus, the cause-effect relationship between GATA3 binding and H3K27me3 marking is still unclear. In summary, the IL-4-driven permissive H4Ac mark at the exon 1A promoter does not obviate the repressive effects of IFN-α that are imposed by the H3K27me3 silencing mark at CNS-1.

Fig. 6.

Positive and negative regulation of the GATA3 gene by IL-4 and IFN-α/β. IL-4 promotes H4 acetylation of the alternative exon 1A promoter region, while IFN-α blocks GATA3 accessibility to the CNS-1 region via H3K27me3 modification. Please refer to the text for further details.

IFN-α/β signaling could inhibit exon 1A expression and promote chromatin modifications within the CNS-1 region by either direct or indirect mechanisms. Perhaps the most linear pathway would involve direct recruitment of STAT2 to the GATA3 locus, which would place STAT2 in the position of a transcriptional repressor. A recent report found that STAT2 could recruit the histone methyl transferase, Ezh2, which is responsible for catalyzing the H3K27 trimethyl modification (36). Alternatively, IFN-α signaling, either through STAT2 or other signaling intermediates, may induce the expression of a downstream repressor. Achieving a complete understanding of GATA3 regulation is difficult due to the complexity of the GATA3 locus. Our data suggest that epigenetic modifications to the CNS-1 region may play an important role in the suppression of GATA3 by IFN-α, and our data suggest that these repressive modifications may block GATA3 binding and autoactivation at this site. Furthermore, there are likely additional regulatory regions involved in the induction of GATA3 by IL-4 and perhaps in the counter-regulation by IFN-α. It is noteworthy that murine CD4⁺ T cells are completely resistant to the counter-regulatory effects of IFN-α/β to block IL-4-driven Th2 development. If the CNS-1 region regulates the suppression of GATA3 in human, but not mouse CD4⁺ T cells, it is likely that small sequence differences within this region may confer species-specific regulation. Alternatively, there may be unique interferon-induced genes that target CNS-1 for silencing in human T cells that are not expressed in mouse. Nonetheless, our data suggest that accessibility of CNS-1 in human CD4⁺ T cells is integral to IFN-α’s inhibitory effect on exon 1A, which would play an important role in driving permanent suppression of Th2 function in response to IFN-α/β.