Long-Range Transcriptional Control of the Il2 Gene by an Intergenic Enhancer

Abstract

Interleukin-2 (IL-2) is a potent cytokine with roles in both immunity and tolerance. Genetic studies in humans and mice demonstrate a role for Il2 in autoimmune disease susceptibility, and for decades the proximal Il2 upstream regulatory region has served as a paradigm of tissue-specific, inducible gene regulation. In this study, we have identified a novel long-range enhancer of the Il2 gene located 83 kb upstream of the transcription start site. This element can potently enhance Il2 transcription in recombinant reporter assays in vitro, and the native region undergoes chromatin remodeling, transcribes a bidirectional enhancer RNA, and loops to physically interact with the Il2 gene in vivo in a CD28-dependent manner in CD4⁺ T cells. This cis regulatory element is evolutionarily conserved and is situated near a human single-nucleotide polymorphism (SNP) associated with multiple autoimmune disorders. These results indicate that the regulatory architecture of the Il2 locus is more complex than previously appreciated and suggest a novel molecular basis for the genetic association of Il2 polymorphism with autoimmune disease.

INTRODUCTION

Interleukin-2 (IL-2) is an important immunoregulatory growth factor that drives T cell and NK cell differentiation, T cell memory, activation-induced lymphocyte apoptosis, and regulatory T cell (Treg) development and homeostasis (1). Genome-wide association studies (GWAS) in humans and genetic studies in mice strongly implicate the Il2 gene in susceptibility to developing Grave's disease, rheumatoid arthritis, celiac disease, multiple sclerosis, psoriasis, Crohn's disease, ulcerative colitis, and type 1 diabetes (T1D) (2).

The Il2 upstream regulatory region has been studied to base-pair resolution, and for 3 decades it has served as a paradigm of tissue-specific, inducible gene transcription (3). T cell receptor (TCR)-induced transcription of the Il2 gene requires an ∼300-bp promoter-enhancer located immediately upstream of the transcriptional start site (TSS) (4, 5), and Il2 transcription is significantly enhanced by CD28 costimulation through a composite NF-κB–AP-1 response element (CD28RE) (6, 7). However, while CD28 augments the transcription of the minimal promoter-enhancer by 5- to 10-fold (6), CD28 costimulation enhances the expression of the endogenous gene by >100-fold (8–10). Also, single-nucleotide polymorphisms (SNP) that associate Il2 with autoimmune disease susceptibility can be found in the ∼100 kb of intergenic space between Il2 and the genetically linked Il21 locus (2). This evidence suggests that additional cis-regulatory elements (CRE) are involved in the regulation of Il2 gene expression and that the Il2 gene has a more extensive regulatory architecture than the proximal promoter-enhancer alone. Furthermore, a locus control region (LCR) has been identified 10 kb upstream of the Il2 gene (11), and Il2 and the neighboring Il21 gene encoding the related cytokine IL-21 are not coexpressed by T helper cells, indicating that these genes inhabit unique chromosomal and transcriptional environments.

We used a four-tiered approach, guided by evolutionary conservation, active chromatin signatures, and 3-dimensional chromosome conformation in a functional interrogation of the intergenic space between Il2 and Il21 for distal CRE. Our studies identify a region ∼80 kb upstream of the Il2 gene (∼20 kb from the Il21 gene) that loops in a CD28-dependent manner to form a long-range contact with the Il2 gene in IL-2-producing T cells. This region contains a functional CTCF-binding insulator and a separate hypersensitive element that binds the p300 coactivator and transcribes a bidirectional, noncoding enhancer RNA (eRNA) in response to TCR/CD28 costimulation. This element is able to enhance transcription from the Il2 promoter-enhancer by 50- to 100-fold in transient reporter assays. These studies reveal that the Il2 locus possesses a more complex transcriptional architecture than previously appreciated and may offer new insight into the genetic underpinnings of autoimmune disease.

MATERIALS AND METHODS

Mice.

C57BL/6 mice, 4 to 6 weeks old, were obtained from The Jackson Laboratory and maintained at the laboratory animal facility of The Children's Hospital of Philadelphia. All animal experiments were conducted according to approved institutional protocols and guidelines.

MAbs and cytokines.

Monoclonal antibodies (MAbs) against CD3ε (2C11; 1 μg/ml), CD28 (37.51; 1 μg/ml), and CTLA4Ig (5 μg/ml) Ab were purchased from BioExpress. IL-2 was purchased from Roche Applied Science. Fluorochrome-conjugated MAbs used for flow cytometry were purchased from BD Biosciences. All molecular biology reagents were analytical grade and purchased from Sigma-Aldrich.

Cell culture.

Single-cell suspensions of spleen and lymph node were prepared. Naive CD4⁺ T cells were isolated by negative selection using CD4 beads (Miltenyi) and stimulated with soluble anti-CD3 and anti-CD28 (1 μg/ml each) for 2 h. Cells were harvested, formaldehyde fixed, and stored at −80°C for downstream application. For the CD4⁺ effector population, CD4⁺ CD25⁻ cells were purified using a Treg purification kit (Miltenyi) and stimulated by phorbol myristate acetate (PMA; 3 ng/ml), ionomycin (1 μM), and 5 U/ml of IL-2. The next day, cells were harvested, washed, and resuspended in complete RPMI with 5 U/ml of IL-2 for an additional 3 days. Cells were harvested, washed, and rested for 4 h and restimulated for 2 h with plate-bound CD3-CD28 antibody. For anergy, CD8⁺ cells were depleted using Miltenyi CD8 microbeads and columns, and the remaining CD4⁺ T cells and antigen-presenting cell were cultured for 24 h with anti-CD3 (1 μg/ml) and CTLA4Ig (5 μg/ml). CD4⁺ T cells were purified using Miltenyi columns and rested in complete RPMI medium for 4 h, followed by restimulation with plate-bound anti-CD3 and anti-CD28 (1 μg/ml each) for 2 h. Supernatants were collected and IL-2 levels were measured by an enzyme-linked immunosorbent assay (ELISA) kit (eBioscience) by following the manufacturer's instructions.

Vectors.

The Il2 promoter-enhancer region (530 bp) and Il21 promoter region (−261 to +17) were amplified by PCR with mouse genomic DNA as a template and cloned into the NheI site in the pGL4.10 basic luciferase vector (Promega Biotech Inc., Madison, WI). Conserved noncoding sequences (CNS) elements were cloned into the KpnI site of the Il2P-pGL4 or Il21P-pGL4 construct. intE-IL-21 (12) was cloned into the KpnI site of the Il21P-pGL4 construct. Control Renilla-luciferase vector was obtained from Promega. The 200-bp region containing the 20-bp consensus CTCF binding site (CBE −78) was cloned into the SacI or SalI site of the CNS-83-IL2PE-pGL4 construct. Mutations were introduced in the CTCF binding site in the CNS-83-CBE-78-IL2PE-pGL4 construct using a site-directed mutagenesis kit (Stratagene). Primers are listed in Table S1 in the supplemental material. CNS kb −83 sequence was analyzed for putative transcription factor binding sites (TFBS) by PROMO analysis (http://alggen.lsi.upc.es/).

Luciferase reporter assay.

EL4.IL-2 cells (5 × 10⁶ cells) were transfected with the plasmids using Lipofectamine according to the manufacturer's instructions and cultured at 37°C for 6 h in complete RPMI medium (supplemented with 10% heat-inactivated fetal calf serum [FCS], 50 μM β-mercaptoethanol, 2 mM l-glutamine, and penicillin-streptomycin antibiotics). Cells were left untreated or were treated with PMA (10 ng/ml; Calbiochem) and ionomycin (1 μg/ml) for 12 to 16 h. 293T cells were transiently transfected with the indicated plasmids using Lipofectamine reagent. After 6 h of transfection, cells were induced with PMA and ionomycin for 24 h. Cells were harvested, lysed, and assessed for luciferase activity. The Dual-Luciferase reporter system (Promega) was used to examine firefly and Renilla luciferase activity. Renilla luciferase was used to normalize transfection efficiency and luciferase activity. Statistical significance between groups was determined using Student's t test (Prism 6).

ChIP-quantitative PCR (qPCR) and ChIP sequencing (ChIP-seq) analysis.

Cells were cross-linked using 1% formaldehyde for 10 min at 37°C and sonicated (Bioruptor; Diagenode), and DNA-protein complexes were isolated using a chromatin immunoprecipitation (ChIP) assay kit (Millipore) according to the manufacturer's instructions with antibodies against CBP-p300 (Santa Cruz), acetylated H3K27 (H3K27ac; Millipore), trimethylated histone H3K27 (H3K27me3; Millipore), trimethylated H3K4 (H3K4me3; Millipore), and CTCF (Millipore). Purified DNA was subjected to quantitative PCR analysis using SYBR green master mix (Applied Biosystems). Primers were designed by Primer Express software (Applied Biosystems) and are listed in Table S1 in the supplemental material. PCR signals from specific or control IgG ChIP reactions are depicted as a percentage of the pre-ChIP input DNA sample. Statistical significance between groups was determined using t test (Prism 6). Alternatively, barcoded libraries were generated from DNA from CD4⁺ T cells subjected to H3K4me3 ChIP and sequenced by the Children's Hospital of Philadelphia Nucleic Acid and Protein Core using the ABI-SOLiD platform, kits, and reagents. Significant H3K4me3 peaks were determined using HOMER with a 10-kb sliding nucleosome model (13). Briefly, raw read distributions separated by ∼150 bp on the plus and negative strands with densities enriched 5-fold compared to those of input control libraries were tagged as significant.

RNA isolation and strand-specific PCR.

Total RNA was extracted from 4 × 10⁶ to 5 × 10⁶ cells using TRIzol (Life Technologies) according to the manufacturer's instructions. Genomic DNA contamination was eliminated by treating total RNA with DNase I. cDNA was transcribed with Superscript III (Life Technologies) with strand-specific antisense or sense primer in accordance with the manufacturer's protocol. PCR was performed with a region-specific primer (nested primer) which was designed to anneal CNS kb −83 (83 kb upstream of Il2) and kb −82.5 regions. Primer sequences are shown in Table S1 in the supplemental material.

FAIRE analysis.

Formaldehyde isolation of regulatory elements (FAIRE) was performed as described previously (14), with slight modifications. Briefly, 5 × 10⁶ cells were cross-linked with 1% formaldehyde for 10 min at room temperature. The same amount of non-cross-linked cells was used as that for the control. Reactions were quenched with 0.125 M glycine, and the cells were rinsed with cold phosphate-buffered saline (PBS) containing protease inhibitors. Cells then were lysed with FAIRE lysis buffer (2% Triton X-100, 1% SDS, 100 mM NaCl, 10 mM Tris-Cl, pH 8.0, 1 mM EDTA buffer, 1× Sigma protease inhibitor cocktail) and sonicated for 10 min (30-s on/off cycles). Lysates then were centrifuged at 20,000 × g for 10 min at 4°C to remove cellular debris, and the supernatant (10%) was collected as the input control and treated with 10 μg proteinase K overnight at 40°C. The remaining lysate was isolated and subjected to three consecutive phenol-chloroform extractions. Each time, the aqueous phase was recovered and mixed with equal volumes of phenol-chloroform. After vortexing, the mixture was centrifuged at 13,000 × g for 5 min and the aqueous phase was recovered. DNA was precipitated with ethanol and pellets were dissolved in Tris-EDTA (TE), and inputs then were incubated overnight at 65°C for reverse cross-linking. DNA was purified using a PCR purification kit (Qiagen) and was quantified using a NanoDrop. Two nanograms of DNA was used for qPCR. Primers used in the reaction are listed in Table S1 in the supplemental material. Statistical significance between groups was determined using Student's t test (Prism 6).

3C analysis.

The chromosome conformation capture (3C) analysis was performed as described previously (15, 16), with minor modifications. Briefly, 1 × 10⁷ cells were fixed with 1% formaldehyde for 10 min at room temperature, and fixation was stopped with 0.125 M glycine. Cells were washed with cold PBS and lysed by cold lysis buffer (10 mM Tris, pH 7.5, 10 mM NaCl, 5 mM MgCl₂, 0.2% NP-40 with protease inhibitor cocktail) for 20 min on ice. The nuclei were harvested and suspended in 1.2× NEB buffer 2 with 0.3% SDS and incubated at 37°C for 1 h. Reaction mixtures were incubated for another hour with 1.8% Triton X-100 at 37°C, followed by overnight incubation with 800 U of HindIII. The reaction was terminated by the addition of a final concentration of 1.3% SDS and incubation at 65°C for 20 min. Samples were further diluted with 1.1× T4 DNA ligase buffer (NEB) containing 1% Triton X-100 and incubated at 37°C for 1 h. T4 DNA ligase (100 U) was added and DNA was ligated at 16°C overnight. DNA was purified by phenol-chloroform extraction. 3C ligation products were quantified in triplicate by quantitative TaqMan real-time PCR using probes and primers (see Table S1 in the supplemental material) which were designed by Primer Express 2.0 software (Applied Biosystems). Digestion efficiencies were monitored by Sybr green qPCR with primer pairs that amplify genomic regions containing or devoid of HindIII digestion sites for the anchor region (+2896 of the Il2 TSS) and at the kb −80 region (see Fig. S1). Bacterial artificial chromosome clone RP23-243E17 (CHORI) containing the entire murine Il2-Il21 locus was digested with HindIII and religated to generate random ligation products of HindIII fragments. The DNA was serially diluted and used to generate a standard curve to which all 3C products were normalized. The 3C signals at the Il2 locus were further normalized to a control locus, Ercc3. 3C products with positive signals were cloned into Topo TA vector (Invitrogen) and sequenced to confirm the presence of ligated bait sequence and distal sequence with the HindIII restriction site. Statistical significance between groups was determined using Student's t test (Prism 6).

RESULTS

Intergenic histone modifications reveal a costimulation-responsive chromatin boundary between the Il2 and Il21 loci.

Chromatin signatures can provide valuable information about the potential regulatory properties of intergenic regions (17), and CRE exhibit a relatively high level of evolutionary sequence conservation compared to intergenic space at large (18, 19). Therefore, we used conserved noncoding sequences (CNS) as landmarks in a preliminary ChIP analysis of active versus repressive histone modifications across the ∼100-kb intergenic region between the Il2 and Il21 loci in a search for potential distal CRE for these genes.

In naive CD4⁺ T cells, the Il2 promoter and upstream regulatory region adopts a closed chromatin conformation with low levels of permissive modifications, such as histone H3/H4 acetylation (10, 20, 21) and H3K4 trimethylation (Fig. 1A, gray), but also low levels of the repressive H3K27 trimethylation mark (Fig. 1A, black). Conversely, the Il21 gene, which is not expressed upon initial activation of naive cells, exhibits very low H3K4me3 levels and a relatively high level of the repressive H3K27me3 marker (Fig. 1A). The intergenic region proximal to Il21 retains this silenced chromatin pattern until reaching a CNS ∼20 kb downstream of Il21 and ∼80 kb upstream of Il2. At this point, H3K27me3 levels drop precipitously and the chromatin exhibits a moderate degree of H3K4 trimethylation (Fig. 1A, asterisk), demarcating a strong chromatin boundary between these two genes. The K27^hiK4^lo chromatin pattern returns on the Il2-proximal side of this boundary, but H3K27me3 drops progressively toward the Il2 gene (Fig. 1A, black).

FIG 1.

Chromatin modifications across the Il2-Il21 multilocus region in primary CD4⁺ T cells. ChIP analysis of H3K4me3 and H3K27me3 marks between the Il2 and Il21 loci, including those for naive (A), TCR-stimulated (B), and TCR/CD28-costimulated (C) CD4⁺ T cells is shown. Promoter-enhancer regions (prom) for each gene are shown, and the percentage of input is indicated on the y axis. Results are the means ± standard errors of the means (SEM) from 3 experiments. The x axis shows the coordinates of the murine Il2-Il21 loci with the Il2 TSS set at 1. Mammalian conservation relative to the Il2 TSS is depicted on the x axis. IL-2 production by TCR- or TCR/CD28-stimulated cells was measured by ELISA (insets in panels B and C). P < 0.01 (**) and P < 0.05 (*) are indicated for statistical comparisons in panels B and C against naive values in panel A.

To further interrogate this region for chromatin elements responsive to signals that regulate the transcription of Il2, we stimulated CD4⁺ T cells through the TCR alone, which results in anergy through active silencing of the Il2 gene (10, 22, 23) or through the TCR together with CD28 costimulation, which induces chromatin remodeling, DNA demethylation, and transcription at the Il2 promoter (10, 24). Stimulation through the TCR alone failed to induce IL-2 production (Fig. 1B, inset) and instead induced an increase in H3K27me3 at the Il2 promoter TSS (Fig. 1B, black). This silenced chromatin signature was not isolated to the promoter but increased approximately 2-fold across the entire Il2-Il21 intergenic region, extending all the way to the Il21 promoter (Fig. 1B, black). H3K27me3 levels also increased at the boundary region (Fig. 1B, asterisk) but remained lower than those at the neighboring regions. The level of H3K4me3 at the boundary in anergic CD4⁺ T cells also decreased compared to that of naive cells (Fig. 1B, gray, asterisk). Costimulation of naive CD4⁺ T cells through the TCR and CD28 led to strong production of IL-2 (Fig. 1C, inset) and induced a remarkable loss of H3K27 trimethylation across the entire Il2-Il21 intergenic region (Fig. 1C, black), indicating that costimulatory signals have large-scale effects on chromatin structure at cytokine genes. However, H3K27me3 levels remained high and H3K4me3 levels remained low at the Il21 promoter (Fig. 1C, gray), consistent with the restricted expression pattern of this gene in select T helper subsets (25). The H3K4me3 modification is a hallmark of active promoters of protein-coding genes, and we observed a strong increase in this mark at the Il2 promoter-TSS in TCR/CD28-costimulated cells (Fig. 1C, gray). However, we also observed a significant peak of H3K4me3 at the kb −80 boundary region in Il2-transcribing cells, increased 2-fold compared to that of naive cells, 4-fold compared to anergic cells, and 2- to 4-fold higher than that at neighboring intergenic regions (Fig. 1C). In addition to active TSS, H3K4me3 is also enriched at many active enhancers (26, 27), suggesting that the kb −80 region contains a transcriptional enhancer for the Il2 gene.

The kb −83 CNS can potently enhance transcription from the proximal Il2 upstream regulatory unit.

To determine whether any of the CNS in the intergenic space between the Il2 and Il21 loci can act as a cis regulatory element, we used a classical, modular promoter-reporter approach. In these assays, individual intergenic CNS were cloned immediately upstream of the bp −500 core Il2 upstream regulatory region (URR) transcriptional unit (4, 5) driving luciferase expression. These recombinant constructs were transiently transfected into EL4 T lymphoblastoid cells, and luciferase activity was measured before and after induction of IL-2 by PMA-ionomycin stimulation. These experiments showed that the kb −83 element is able to enhance transcription of the core Il2 upstream regulatory region by 50-fold or greater (Fig. 2A; also see Fig. S2 in the supplemental material), while none of the other intergenic CNS tested at kb −3, −31, −35, −42, −62, −63, or −80 were able to augment Il2 transcription. The activity of the kb −83 CNS was specific for Il2, as this region was not able to enhance transcription from the Il21 promoter (Fig. 2B). Instead, the nearby kb −80 CNS exhibited moderate enhancer activity for the Il21 promoter (Fig. 2B).

FIG 2.

Functional interrogation of intergenic CNS for enhancer activity using recombinant Il2 promoter-reporter constructs. (A) Individual CNS from the Il2-Il21 intergenic region were cloned upstream of a 500-bp Il2 promoter-upstream regulatory region (Il2P-URR) driving firefly luciferase. Each construct was cotransfected together with a constitutive Renilla control in EL4.IL-2 cells, and luciferase activity was measured 16 h after stimulation with PMA-ionomycin. The y axis depicts the ratio of firefly luciferase activity from each CNS normalized against a promoterless control construct (pGL4). All samples also were normalized against Renilla luciferase activity as a loading control. (B) The capacity of the kb −80 and kb −83 CNS to enhance transcription from the Il21 promoter was assessed as described for panel A. The Il21 internal enhancer (intE) was used as a positive control. (C) Prediction of transcription factor binding sites (PROMO) within the kb −83 CNS. (D) The entire 403-bp −83 CNS, a 205-bp 5′ fragment containing one NFAT and the first four AP1 sites, or a 226-bp 3′ fragment containing NFAT, STAT, AP1, and p300 sites was cloned adjacent to the Il2 promoter-enhancer-reporter construct and cotransfected with NFAT, p300, or c-Rel expression constructs in 293T cells or EL4.IL-2 cells. Luciferase activity was measured as described for panel A. (E) kb −83 CNS was cloned into the promoterless pGL4 backbone (row 2) or 5′ to the Il2 promoter-enhancer (Il2PE) in forward (row 3) or reverse (row 5) orientation. A 200-bp DNA fragment containing the kb −78 element with the CTCF binding site (asterisk) ablated by site-directed mutagenesis was cloned between the kb −83 CNS enhancer and the Il2P-URR (row 4). Constructs were transfected in EL4.IL-2 cells, and luciferase activity was measured as described for panel A. P < 0.0001 (****), P < 0.01 (**), and P < 0.05 (*) are indicated to denote significance between tests and Il2P-luc controls.

Analysis of the primary nucleotide sequence of the −83 CNS reveals an array of putative binding sites for factors known to regulate Il2 gene expression, including AP-1 (Fos/Jun), p300, NFAT, and other Rel family members, such as NF-κB (Fig. 2C). NFAT, AP-1, and the NF-κB family member c-Rel cooperate with the coactivator p300/CBP to trans-activate Il2 transcription (28–31). The transformed epithelial cell line 293T does not express NFAT or c-Rel and cannot drive the transcription of the core Il2 upstream regulatory unit (Fig. 2D). Ectopic expression of c-Rel or NFAT together with p300 rescued the transcription of the core Il2 upstream regulatory region to a moderate degree (Fig. 2D). However, cotransfection of these factors with the core upstream regulatory unit paired with the kb −83 module led to a marked increase in transcription (Fig. 2D), indicating that, like the proximal Il2 enhancer (29), the kb −83 CNS is an NFAT-, p300-, and c-Rel-responsive enhancer. As an initial step to determine the minimal region required for enhancer function, we split the 403-bp kb −83 CNS region into two halves and cloned them upstream of the Il2 promoter-enhancer module. We found that ∼90% of the enhancer activity in 293T cells is confined to the 5′ half of the CNS (Fig. 2D), where a majority of the consensus elements for NFAT and AP-1 are clustered (Fig. 2C). However, the kb −83 CNS sequence was able to drive much stronger transcription from the core Il2 upstream regulatory module in EL4 T lymphoblastoid cells without the need for ectopic NFAT or c-Rel expression (Fig. 2D). The 5′ half of the kb −83 CNS was able to function to some extent on its own, but full enhancer activity required the entire sequence in these cells (Fig. 2D). This suggests that additional T cell-specific factors that bind to elements in the 3′ half cooperate with NFAT, AP-1, and c-Rel to achieve full activity.

Enhancers are classically defined in recombinant reporter assays as sequences that lack intrinsic promoter activity but can enhance transcription from a defined promoter element in a position- and orientation-independent manner. We confirmed that the kb −83 CNS does not display intrinsic promoter activity (Fig. 2E, row 2), and insertion of a 200-bp functionally inactive DNA fragment between the kb −83 CNS and the core Il2 upstream regulatory unit did not significantly affect its activity (Fig. 2E, row 4), suggesting that this element is position independent. However, in multiple experiments we observe that reversing the orientation of this element with respect to the Il2 upstream regulatory unit largely abrogates its activity (Fig. 2E, row 5), indicating that the kb −83 enhancer collaborates with the Il2 upstream regulatory region in a specific orientation. Interestingly, an original study of the Il2 upstream regulatory enhancer similarly showed a loss of ∼75% of its activity when its orientation was reversed (5), and a more recent and detailed study found that the Il2 upstream regulatory element is particularly sensitive to cis-element spacing and orientation (32). These data show that the distal kb −83 CNS element can potently enhance transcription from the core Il2 upstream regulatory unit if these two elements are artificially brought into proximity, and this element shares certain structural constraints with the Il2 upstream regulatory region that appear to render its activity specific for the Il2 gene.

Coregulated chromatin remodeling and coactivator recruitment at the endogenous Il2 promoter and kb −83 distal enhancer elements upon activation.

Stimulation through the TCR and CD28 is accompanied by posttranslational modification of histones and remodeling of a positioned nucleosome within the Il2 upstream regulatory region (10, 20, 29, 30, 33, 34). This can be observed in recent DNase I hypersensitivity mapping data sets from the ENCODE project (18). Cell lineages that are incapable of producing IL-2 (e.g., kidney, B cells, and Treg) exhibit an inaccessible chromatin profile at the URR (Fig. 3A), while resting CD4⁺ T cells showed slight accessibility at the Il2 TSS (Fig. 3A, green). This may represent poising of the locus in lineages with IL-2-producing potential or may be due to the presence of memory cells in the unfractionated CD4⁺ populations used in these studies. The kb −78 region contains a strong hypersensitivity site in all cell types (Fig. 3B), while the kb −80 CNS exhibits constitutive hypersensitivity in the T cell lineage only (Fig. 3B). Conversely, the kb −83 distal enhancer exhibits a closed chromatin conformation in non-IL-2-producing cells (Fig. 3B), a closed conformation in unstimulated EL4 cells as measured by FAIRE (Fig. 3D, left, green), and a slightly poised conformation in resting CD4⁺ T cells (Fig. 3B, green), which is similar to the pattern exhibited by the Il2 promoter-URR (Fig. 3A).

FIG 3.

Inducible chromatin remodeling ∼80 kb upstream of Il2. (A and B) High-resolution DNase I hypersensitivity analysis of the Il2P-URR (A) and the kb −78 to kb −83 region (B) in murine activated CD4⁺ T cells, resting CD4⁺ T cells, activated regulatory T cells (Treg), B cells, and kidney tissue from the ENCODE project. Mammalian conservation (cons) is shown in blue (also in panels E and F). mu, murine; chr, chromosome. (C and D) FAIRE and ChIP analysis of chromatin accessibility and p300 occupancy at the Il2P-URR (C) and the kb −78 to kb −83 region (D) in resting (green) versus stimulated (orange) EL4.IL-2 cells (FAIRE) or CD4⁺ T cells (p300 ChIP). The y axis depicts the percentage of input for ChIP and fold over input for FAIRE. Means ± SEM from 3 experiments are depicted. (E) ChIP-seq analysis of H3K4 trimethylated chromatin at the Il2 TSS (left) and across the region 78 kb to 83 kb upstream of the Il2 TSS (right) in resting versus TCR/CD28-activated murine CD4⁺ T cells from this study. Histograms show filtered peaks with read densities of >4-fold over background levels. (F to H) H3K27ac ChIP-seq peaks (F; orange), chromosome segmentation analysis by hidden Markov modeling (G; ChromoHMM), and transcription factor ChIP-seq binding (H) in human B lymphoblastoid cell lines from the ENCODE project are shown. hu, human. (I) CTCF occupancy (left) was assessed in resting (green) versus activated (orange) CD4⁺ T cells by ChIP-qPCR at the indicated regions upstream of Il2 that contain CTCF consensus binding sequences. On the right, the kb −83 CNS was cloned 5′ of the Il2P-URR (row 2). A 200-bp DNA fragment containing the CTCF consensus binding sequence at kb −78 was cloned between (row 3) or adjacent to (row 4) the CNS-83 enhancer and the Il2P-URR. The CTCF binding site (asterisk) was ablated by site-directed mutagenesis (row 5). Constructs were transfected in EL4.IL-2 cells, and luciferase activity was measured as described for panel A. P < 0.001 (***), P < 0.01 (**), and P < 0.05 (*) are indicated to denote significance between tests and controls. Statistical analysis could not be performed on UW ENCODE DNase I tracks (n = 1).

Upon activation, CD4⁺ T cells exhibit a stronger accessibility signal around the core promoter/TSS and also at the URR/enhancer (Fig. 3A, orange). The same accessibility shift is observed between resting and stimulated EL4 cells as that measured by FAIRE (Fig. 3C, left, orange) and is accompanied in CD4⁺ T cells by increased p300 occupancy at the URR (Fig. 3C, right, orange), as observed previously (29). Under the same conditions, the kb −83 CNS also exhibits a stronger accessibility signal in both CD4⁺ T cells (Fig. 3B, orange) and EL4 cells (Fig. 3D, left, orange), indicating that the nucleosomes at these regions have been displaced by DNA binding protein complexes. Similar to the promoter-URR, activation-induced chromatin remodeling at the kb −83 CNS region also was accompanied by increased p300 occupancy (Fig. 3D, right). We also conducted a high-resolution, ChIP-seq analysis of H3K4 trimethylation across the kb −78 to kb −83 region upstream of the Il2 locus in murine CD4⁺ T cells. This modification marks genomic regions undergoing active transcription and is found at the promoters of expressed genes, but it also marks distal enhancer elements engaged in looping interactions with transcriptionally active genes. As expected, resting CD4⁺ T cells exhibit no H3K4me3 at the Il2 promoter (Fig. 3E, left, green), while costimulation through the TCR and CD28 led to strong H3K4me3 modification at the promoter (Fig. 3E, left, orange), coincident with induction of Il2 transcription (Fig. 1). Importantly, CD4⁺ T cells exhibited coordinate regulation of chromatin structure at the kb −83 element. While unstimulated cells showed little or no enrichment of the H3K4me3 mark across the entire region (Fig. 3E, right, green), stimulation through the TCR and CD28 resulted in strong accumulation of the H3K4me3 mark on the nucleosomes neighboring the occupied kb −83 element (Fig. 3E, right, orange) as well as at the kb −78 and kb −80 elements (Fig. 3E, right, orange). This is consistent with our lower-resolution ChIP-qPCR survey of the H3K4me3 mark at this region (Fig. 1). Importantly, these chromatin dynamics at the kb −83 element are conserved between mouse and human, as various human lymphoblastoid cells exhibit strong H3K27 acetylation, a mark of active enhancers (Fig. 3F, orange). Indeed, hidden Markov modeling-based chromosome segmentation analysis (chromoHMM) from the ENCODE project predicts this region to be a strong enhancer element in human lymphocytes (Fig. 3G, orange) based upon the histone modifications present and on the fact that this element is bound by NFAT, AP-1 (c-Fos/c-Jun heterodimers), Runx1, and p300 in human lymphoblastoid cells (Fig. 3H).

The kb −78 region binds CTCF and exhibits insulator activity in vitro.

Our ChIP-qPCR survey (Fig. 1) showed that a region ∼78 kb upstream of the Il2 gene exhibits lower levels of H3K27 trimethylation than neighboring intergenic domains. In the mammalian genome, CTCF-cohesin complexes serve multiple functions based upon their ability to define distinct chromatin domains. CTCF and cohesin can constitute a barrier that prevents spreading of neighboring chromatin structures into genomic regions, can inhibit gene expression by promoting heterochromatin formation and preventing communication between genes and distal regulatory elements (DRE), or can promote gene expression by mediating active looping between DRE and promoters (35). The kb −78 Il2 element is bound by CTCF-cohesin complexes in a number of human cell lines analyzed as part of the ENCODE project (Fig. 3H, blue) and is predicted by hidden Markov modeling-based chromosome segmentation analysis (18) to be a CTCF boundary element (Fig. 3G, blue). Similarly, we find that the kb −78 element is occupied by CTCF in naive and activated CD4⁺ T cells in our murine system (Fig. 3I, left), while several other intergenic CTCF consensus sequences upstream of the Il2 gene were not bound by CTCF in these cells (Fig. 3I, left).

To determine whether the kb −78 boundary region encodes functional barrier activity, we cloned an ∼200-bp DNA sequence containing the CTCF binding element between the kb −83 enhancer and the core Il2 promoter-enhancer unit in a luciferase reporter construct. Insertion of the kb −78 CTCF binding region resulted in an ∼5-fold reduction in the capacity of the kb −83 CNS to enhance transcription from the proximal Il2 promoter-enhancer (Fig. 3I, row 3), as would be predicted if this element had insulator properties. The barrier activity of the kb −78 region was position dependent, as situating this element upstream of the kb −83 enhancer-promoter unit had no effect on transcription (Fig. 3I, row 4). Likewise, barrier activity was sequence specific and dependent upon an intact CTCF binding sequence, as a 5-nucleotide core mutation abrogated the enhancer blocking effect (Fig. 3I, row 5). These results, together with the chromatin mapping data in Fig. 1, indicate that the kb −78 region is a native CTCF-bound element that demarcates a transition between regions with distinct chromatin marks in vivo and has the characteristic enhancer-blocking activity of an insulator in vitro. This element could mediate looping between the kb −83 enhancer and the Il2 gene.

The native kb −83 CNS transcribes an inducible eRNA and loops to interact with the Il2 promoter in situ in IL-2-producing cells.

Our data show that the distal kb −83 CNS element can potently enhance transcription from the core Il2 promoter-enhancer if these elements are artificially taken from their native genomic contexts and brought into physical proximity in recombinant reporter constructs. In the native genomic environment, cis-regulatory elements enhance gene transcription by looping to physically interact with and recruit coactivator complexes to distant promoters (36). This in situ enhancer activity results in inducible transcription not only of the target gene but also of bidirectional, noncoding RNA (ncRNA) encoded by the active enhancer (37). Genome-wide screens for intergenic, bidirectional enhancer RNAs (eRNAs) recently have been used to successfully identify and validate active, tissue-specific enhancers in various cell types (19). Therefore, the combination of inducible coactivator occupancy, inducible promoter looping, and inducible eRNA transcription represents a potent strategy to identify active, stimulus-responsive enhancers of tissue-specific genes like Il2 in the context of their native genomic environments.

We show that the kb −83 CNS region undergoes inducible chromatin remodeling (Fig. 3B and D) and increased p300 occupancy (Fig. 3D) in response to stimuli that induce Il2 transcription, as would be expected from an endogenous enhancer of the Il2 gene. In addition, we find that these same signals induce the trimethylation of H3K4 at kb −83 (Fig. 3E), suggesting active transcription in this region. To determine whether the kb −83 CNS transcribes eRNA in response to signals that induce transcription of the Il2 gene, primary CD4⁺ T cells were stimulated through the TCR alone or costimulated through the TCR and CD28, and cDNA was generated using primers specific for genomic regions surrounding the kb −83 enhancer. Reverse transcription-PCR (RT-PCR) analysis revealed the presence of noncoding enhancer RNA derived from both strands of the CNS kb −83 region in TCR/CD28 costimulated T cells, while these transcripts were absent from or greatly reduced in T cells in which anergy was induced by TCR ligation alone (Fig. 4A). The same eRNA also was detected in PMA-ionomycin-stimulated EL4 cells but not in unstimulated EL4 cells. In contrast, no specific transcript was detected under any condition at the neighboring kb −82.5 region in either cell type (Fig. 4B). We were unable to detect this species from oligo(dT)-primed cDNA, indicating that the detected transcripts lack a poly(A) tail, which is consistent with the generally nonpolyadenylated nature of eRNAs (38). These data show that the inducible transcription of the Il2 gene is associated with the transcription of noncoding eRNA from the kb −83 CNS in a manner consistent with an active, endogenous enhancer.

FIG 4.

kb −83 CNS transcribes an inducible, bidirectional, noncoding enhancer RNA. Strand-specific RT-PCR was performed using RNA from TCR- versus TCR/CD28-activated primary CD4⁺ T cells (A) or unstimulated versus PMA-ionomycin-stimulated (P/I) EL4.IL-2 cells (B). Bidirectional transcripts from the kb −83 CNS (top) or a kb −82.5 control region (bottom; note some nonspecific amplification in the sense reaction for this region in panel A) were detected using antisense-specific (AS) or sense-specific (S) primers for cDNA synthesis. Oligo(dT) was used to detect poly(A) RNA. Genomic DNA was used as a positive control for PCR amplification. No reverse transcriptase (RT) control reactions were used to rule out the contamination of RNA with genomic DNA.

Distal enhancer elements regulate transcription in native chromosomal environments by looping to physically interact with distant genes (36). The formation of these inducible, higher-order chromatin structures is closely associated with and, in some cases, dependent upon the transcription of eRNA at the distal enhancer (37). To determine if TCR/CD28-dependent transcription of the Il2 gene and the kb −83 CNS is accompanied by a topological interaction between these two distant regions, we utilized a chromosome conformation capture (3C) approach. Using a primer-probe set within a restriction fragment in the second intron of the Il2 gene as the bait, we used quantitative PCR to measure the efficiency at which distal intergenic regions were captured in proximity to the Il2 gene during chemical cross-linking of native chromatin from T cells stimulated through the TCR versus the TCR and CD28. In anergic CD4⁺ T cells, we found a strong looping interaction between the kb −10 LCR and the Il2 gene, but no significant interaction between the Il2 gene and the kb −78 to −83 regulatory region was detected (Fig. 5, peaks). Costimulation through the TCR and CD28 resulted in the induction of efficient looping between the Il2 gene and the kb −78 to −83 regulatory region at the expense of the LCR-Il2 gene interaction (Fig. 5, black lines). An interaction between the kb −78 to −83 region and the endogenous Il2 gene also was observed in stimulated (Fig. 5, black lines) but not unstimulated (Fig. 5, gray lines) EL4 cells. This CD28-dependent looping was specific for the region including the distal boundary and enhancer elements, as a more proximal region at kb −35 showed no significant interaction with the Il2 gene under stimulatory conditions in either cell type (Fig. 5). These data demonstrate that the chromatin containing the endogenous kb −83 distal CRE undergoes a CD28-dependent topological change to interact with the Il2 gene in T cells under conditions that induce Il2 transcription. Together with our data showing that the kb −83 CNS can potently enhance Il2 transcription in the context of recombinant reporter constructs, the fact that this region undergoes a shift in accessibility, recruits coactivator complexes, transcribes a bidirectional eRNA, and loops to interact with the Il2 gene specifically in response to TCR/CD28 costimulation strongly implicates the −83 CNS as an active enhancer of Il2 gene transcription in the context of the endogenous Il2 locus.

FIG 5.

Three-dimensional chromosome capture (3C) analysis of long-range promoter-enhancer looping at the Il2 locus. Looping between the Il2 promoter and the LCR (−10 kb) or the kb −83 enhancer was measured by 3C in primary CD4⁺ T cells (A) stimulated through the TCR or the TCR and CD28 or in resting (unstimulated) versus PMA-ionomycin-stimulated EL4.IL-2 cells (B). CNS are indicated by peaks, and a HindIII map of the depicted region is shown. The y axis depicts the relative cross-linking efficiency (means ± SEM) between the promoter and the LCR or the kb −83 enhancer, calculated as described in Materials and Methods. Cross-linking efficiency at kb −35 is shown in panel A as a negative control for chromatin looping. P < 0.01 (**) and P < 0.05 (*) are indicated for significant differences between TCR and TCR plus CD28 in panel A and stimulated versus unstimulated in panel B.

DISCUSSION

IL-2 is a potent cytokine that promotes NK cell proliferation and function and effector T cell differentiation into Th1 and Th2 lineages, and it inhibits Th17 and T follicular helper (Tfh) differentiation (1, 39). IL-2 also is required for the development and homeostasis of regulatory T cells (Treg), which suppress inflammation and mediate tolerance (40). Il2 gene expression is under tight developmental and contextual control, and full induction in T cells requires signals from both the TCR and a costimulatory receptor, such as CD28 (3, 30, 41). A definition of the full regulatory architecture of the Il2 gene is important to further understand its diverse roles in immunity, tolerance, and disease pathogenesis.

In the current study, we find that the transcriptional architecture of the Il2 locus incorporates a distal regulatory region 78 to 83 kb upstream, containing a boundary element and an enhancer that loop to interact with the Il2 gene upon TCR/CD28 costimulation. Thus, Il2 joins genes like the erythroid-cell-specific β–globin, the T helper 2-specific Il4-Il5, the T helper 1-specific Ifng-Il26, and the T helper 17-specific Il17a-Il17f gene clusters that are distributed over hundreds of kilobases and contain multiple enhancers, locus-control regions, insulators, and silencers (42–47). The kb −78 element within this regulatory region binds CTCF in vivo, exhibits in vitro enhancer blocking activity in transcriptional reporter assays, and exhibits a unique pattern of histone methylation that may help to establish the mutually exclusive expression pattern exhibited by Il2 and Il21. The kb −83 element within this region exhibits the accessible signature of an active enhancer (42) and is bound by p300/CBP in T cells and multiple lymphoid cell lines. This element potently enhances transcription from the proximal promoter/enhancer in reporter assays, and in its native chromatin environment it acquires H3K4 trimethylation and transcribes a bidirectional eRNA in a TCR-induced, CD28-dependent manner in IL-2-producing CD4⁺ T cells.

We detected looping between the Il2 gene and both the kb −10 LCR and the kb −80 regulatory region. The LCR-promoter loop was detectable in both unstimulated and TCR/PMA-stimulated EL4 and CD4⁺ T cells, consistent with the ability of this element to establish the Il2 promoter-enhancer as an independent transcriptional unit in a variety of chromatin contexts. However, the interaction between the distal region and the gene did not occur in naive/unstimulated cells and was induced in CD4⁺ T cells only by costimulation through the TCR and CD28. Long-range CRE-promoter interactions commonly involve the chromosome architectural proteins CTCF and cohesin, which are known to contribute to promoter-enhancer looping at the Ifng and Il17 cytokine loci (16, 47–49). Signal-induced transcripts encoded by the kb −83 enhancer tracked with induction of coding transcripts from the Il2 gene, which is a hallmark of active distal enhancers and eRNAs (19, 37). eRNAs are enriched at enhancers engaged in inducible looping interactions and have been shown to actively stabilize inducible enhancer-promoter looping and recruitment of coactivators and RNA polymerase II to active promoters by binding to components of the Mediator and cohesin complexes (50–52). Based on our results, we propose a model in which the kb −83 eRNA cooperates with the CTCF element at kb −78 to induce and/or stabilize a long-range interaction with the kb −80 intergenic regulatory region, bringing the kb −83 enhancer in proximity to the Il2 promoter (Fig. 6). Interestingly, the activity of this element in our promoter-reporter assays is highly dependent upon its orientation with respect to the Il2 upstream regulatory region. Previous studies have shown that the proximal Il2 enhancer likewise is sensitive to cis-element spacing and orientation. These results suggest to us that the kb −83 enhancer is not a generic cis-regulatory element but rather loops to interact with the upstream Il2 regulatory region in a specific orientation that matches the architectural constraints characteristic of the proximal Il2 enhancer. This also may explain our observation that the kb −83 CNS does not act as an enhancer for the Il21 promoter.

FIG 6.

Long-range regulation of Il2 transcription by the intergenic −78/−83 boundary/enhancer elements. The model shows CD28-dependent eRNA transcription and looping between the Il2 promoter and distal regulatory elements identified in this study.

Our results indicate that TCR/CD28-costimulated CD4⁺ T cells utilize a distal enhancer for high-level expression of Il2 and suggest a novel level of tissue-specific control of this gene. CD8⁺ cytotoxic T lymphocytes, anergic cells, and regulatory T cells are poor IL-2 producers, even when stimulated through the TCR and CD28, because these lineages actively silence Il2 gene expression (41). Our studies indicate that the epigenetic silencing mechanisms known to operate at the level of the proximal promoter-enhancer in anergic T cells also extend over the entire 100-kb intergenic space between the Il2 and Il21 genes, including the kb −83 enhancer. These findings have significance in the context of the genetic basis of disease, as the Il2 gene is implicated in multiple autoimmune disorders in both humans and mice (2). In the murine NOD genetic model of type 1 diabetes, the protective B6 Idd3 allele of Il2 confers increased expression in both CD4⁺ T cells and dendritic cells (53, 54), but the causative polymorphism within this haplotype is not known. Dozens of SNPs exist within the kb −80 intergenic regulatory region in both the human and mouse genomes, but to date none of these polymorphisms have been linked to disease susceptibility in mice. Given that the majority of disease-associated SNPs are located outside gene coding regions (55), it is possible that sequence polymorphisms within the kb −78 boundary element, the kb −83 enhancer, or other unidentified intergenic CRE are responsible for the differences in Il2 expression between the NOD and the B6 alleles. Indeed, human GWAS studies have associated a common Il2/Il21 intergenic polymorphism (rs6822844) with susceptibility to rheumatoid arthritis, T1D, inflammatory bowel disease, celiac disease, lupus, and multiple sclerosis (56–59), and this SNP maps within a few kilobases of the kb −80 regulatory region identified in this study. Because genome-wide association studies are only powered to identify groups of SNPs within a relatively large haplotype block, not individual causal SNPs, these studies implicate the distal boundary element and/or the enhancer as potential contributors to genetic susceptibility to autoimmune disease in humans.