IL-2 REGULATES EXPRESSION OF C-MAF IN HUMAN CD4 T CELLS¹

PBMC and CD4 T cell isolation, culture and treatment

PMBCs were isolated from buffy coats from healthy donors (purchased from National Blood Service, Tooting, UK) by density gradient centrifugation using LSM 1077 (PAA laboratories). CD4 T cells were purified from PBMCs by positive selection with human CD4 microbeads (Miltenyi Biotec) according to manufacturer’s instructions. The purity of CD4 T cells ranged from 93% to 96% by FACS analysis using FITC-conjugated αCD3 and PE-conjugated αCD4 mAbs (both from DAKO) on a FACScalibur instrument (Becton Dickinson). Cells were maintained in culture medium (CM) of RPMI 1640 medium (PAA Laboratories, Austria) supplemented with 10% fetal bovine serum (FBS, PAA laboratories), 2mM L-glutamine and penicillin-streptomycin (100 IU/ml and 100μg/ml, both from Sigma-Aldrich, UK). When indicated, CD4 T cells were stimulated by 2 μg/ml PHA-L (EY laboratories, CA) for 72h and rested in CM for ≥16 hours prior to activation with 100U/ml IL-2 (Roche Diagnostics). CD4 T cells were activated via the TCR by stimulation with plate-bound αCD3 (UCHT1, 10 μg/ml) and soluble αCD28 antibody (2 μg/ml; both from Ancell) with or without the JAK3 inhibitor R333 (5μM; a generous gift from Rigel Pharamceuticals, CA). IL-6 (R&D Systems) was used at 10 ng/ml for stimulation experiments. CD4 T cells were inhibited by treatment with anti-human IL-2 antibody (10 μg/ml; Biosource Inc.) or HAT (daclizumab, 10 μg/ml; kindly provided by Hoffmann-La Roche, NJ), respectively.

Chromatin immunoprecipitation (ChIP)-cloning and ChIP-seq

ChIP-cloning experiments were performed using 1.5 × 10⁷ CD4 T cells stimulated with IL-2 for 20 minutes followed by a 10 minute crosslink with formaldehyde (0.37%) at 37 °C, as previously described (25). The sonicated chromatin was immunoprecipitated for an hour using μMACS Protein-A microbeads (Miltenyi Biotec) and anti-Stat5a or anti-Stat5b monoclonal or control mouse IgG antibodies (all from Zymed). Purified chromatin was blunt-end repaired using T4 DNA polymerase (NEB), ligated to linkers, PCR amplified, and cloned into TOPO vector (Invitrogen) and sequenced (Cogenics).

For identification of the sequences enriched in regions associated with specific histone phenotypes (ChIP-seq experiments), ChIP was performed on native chromatin from ex vivo differentiated human TH2 cells (26). Chromatin was prepared by using the protocol of Feil and colleagues with some minor modifications (http://www.epigenome-noe.net/researchtools/protocol.php?protid=2). Briefly, nuclei were treated with micrococcal nuclease (10U/μl, 7 min) and mono and dinucleosomal chromatin was recovered. Chromatin quality was assessed by agarose gel electrophoresis and semi-quantitated using a nanodrop. Chromatin (20μg) was immunoprecipitated with 3μg anti-H3K4me3 (ab8580) or anti-IgG control, and magnetic Protein G beads (Active Motif). Prior to sequencing, the diversity of sequences enriched by ChIP was assessed by comparing qPCR of input DNA with anti-IgG or anti-H3K4me3 immunoprecipitated DNA using multiple probe sets (ChIP-qPCR). ChIP libraries were prepared from amplified DNA (17 cycles) from two biological replicates of TH2 cells according to the manufacturer’s protocol (Illumina). DNA from separate experiments was sequenced on GAII genome analyzers (BRC Core Facility KCL). Detection of regions of enriched H3K4me3 relative to input DNA was performed by using MACS and SICER programs. Output was converted to browser extensible data (BED) files displaying the number of Tags in 200bp windows and visualized within the UCSC browser.

Preparation of cell extracts and Western blotting

Nuclear and total cell extracts were prepared as described previously (27). Samples were resolved on 8% SDS–PAGE gels and transferred to Immobilon PVDF membrane (Millipore). Western blots were performed using pan-STAT5 (Invitrogen), anti-phospho-STAT5Y694 (New England Biolabs), c-maf polyclonal antibody (M-153, Santa Cruz Biotechnology), anti-Lamin-B or anti-tubulin antibodies (both from Invitrogen) and developed using ECL Plus chemiluminescence reagent (GE Healthcare).

RNA Isolation and Quantitative real time PCR

Total RNA was extracted using TRIzol Reagent (Invitrogen). RNA (250 ng) was reverse-transcribed to cDNA using RevertAid-H Minus M-MuLV Reverse Transcriptase (Fermentas) following manufacturer’s instructions. qRT-PCR was performed in triplicate using 2X Precision SYBR Green MasterMix (PrimerDesign Ltd, UK) in a total reaction volume of 20 μl using the ABI Prism 7900 HT Sequence Detection System (Applied Biosystems) although ChIP-qPCR reactions were performed in duplicate. The human PPIA or 18S rRNA genes were used as reference controls for the various experiments. Ct values were exported with the sequence detector SDS 2.3 software (ABI). Gene-specific primers for PPIA, 18S rRNA, C-MAF, CD25 and IL-2 were obtained from PrimerDesign Ltd. UK. The sequences of all other primers are given in Figure S3.

Luciferase Reporter Assay

HEK293T cells were cultured and transfected to reconstitute IL-2R signaling using the calcium phosphate method, as previously described (28). Trimerized GAS1-4 sequences were synthesized (MWG), cloned into pGL4.23 vector (Promega Corp.), and verified by sequencing (Cogenics). Luciferase assays were performed by using the Dual-Glo luciferase Kit (Promega Corp.), according to manufacturer’s instructions. Firefly luciferase activities were standardized to the corresponding Renilla luciferase activities for each sample to control for transfection efficiencies.

TH2 skewing and Intracellular cytokine staining

Fresh cord blood units (CBU) were obtained from the Programa Concordia Banc de Sang i Teixits, Barcelona, Spain, with prior consent and ethical committee approval. CBU were diluted 1:1 with RPMI 1640 (Lonza) supplemented with 0.6% trisodium citrate and 40 nM 2-ME. Cord blood mononuclear cells were separated by density gradient centrifugation using Ficoll-Paque Premium (GE Healthcare, UK). CD4 T cells were isolated by negative selection using a Miltenyi isolation kit. Mean (±s.d.) positivity for CD45RA on isolated CD4 cells was 93% ± 5% (n=3). Cord blood CD4 cells (0.5-1 million) were activated with IL-2 as follows: αCD3 (10 μg/ml), αCD28 (2 μg/ml; both BD) and IL-2 (30U/ml) ± HAT (10 μg/ml). TH2 skewing was performed as previously described (26) 0.5-1 million cord blood CD4 cells stimulated with αCD3 (10 μg/ml), αCD28 (2 μg/ml), IL-2 (30U/ml), IL-4 (12.5ng/ml; NBS Biologics), anti-IFN-γ and anti-IL-10 (both 5μg/ml, BD) ± HAT (10 μg/ml).

Intracellular staining for IL-4, and IFN-γ was carried out according to standard methods. Briefly, cultured cells were restimulated with Phorbol Myristate Acetate (PMA; 50ng/ml) and Ionomycin (1mM, both Sigma) for 4.5h at 37°C in the presence of Brefeldin A (3 μg/ml; eBioscience) in 1ml of CM. Washed cells were fixed with 4% PFA, blocked with 50% FCS (v/v in PBS), permeabilized with 0.1% saponin (w/v in 1% FCS in PBS) and stained for 30 min in the dark at 4°C with monoclonal antibodies against IL-4 (BD) and IFN-γ (Caltag) or appropriate isotype controls (mouse IgG1-PE (BD) and mouse IgG1-APC (Caltag)) at recommended concentrations. Within 4h of staining, data from live cells, based on forward/side scatter profiles, were acquired on a FACSCalibur flow cytometer (CellQuest software; BD) and analyzed with FlowJo (Treestar Inc, Ashland, OR).

Data analysis

Data analysis utilized Microsoft Excel (Microsoft Corporation, USA) and GraphPad Prism v4 (GraphPad Software, Inc.). Statistical analysis was performed by using paired parametric and non-parametric tests as appropriate. p values of < 0.05 were considered to be statistically significant.

IL-2 activated STAT5 binds to the C-MAF locus in vivo in CD4 T cells

IL-2 is one of the earliest cytokines produced by activated T cells and mediates its actions primarily through the activation of STAT5 proteins. A STAT5-chromatin immunoprecipitation assay (ChIP) was performed using chromatin from freshly isolated CD4 T cells to identify in vivo IL-2-activated STAT5 gene targets. The immunoprecipitated chromatin yielded a number of distinct clones based on sequencing. One clone mapped to chromosome 16 at −152,916 to −153,096 upstream of the C-MAF gene, and contained a consensus GAS motif (Figure 1A). Since IL-2 and STAT5 play significant roles in gene regulation during TH2 cell differentiation, we examined if IL-2-induced STAT5 also regulates C-MAF gene expression.

In vivo binding of STAT5 was assessed in cross-linked chromatin fragments derived from either IL-2-stimulated or untreated freshly isolated CD4 T cells by immunoprecipitation with anti-STAT5 antibodies in a ChIP assay. DNA was isolated from precipitated chromatin and assayed for the presence of the upstream C-MAF sequence by using PCR with specific primers that detect the aforementioned cloned fragment. IL-2 treatment induced binding of STAT5a and STAT5b to this sequence (Figure 1B). As expected, this sequence was not enriched in the IL-2-treated sample by ChIP performed with control IgG (Figure 1B). Thus, there was specific IL-2-induced binding of STAT5 proteins to a sequence upstream of the C-MAF transcriptional start site (TSS). Since this sequence is located at a long distance upstream of the TSS of the C-MAF gene, we next determined if IL-2 induced STAT5 binds to other 5′ upstream regions that were closer to the C-MAF promoter. We postulated that transcriptionally relevant binding sites would be located within cell-type specific DNase1 hypersensitive (DHS) sites (29, 30).

The genome-wide DHS study of CD4 T cells identified three DHS sites (DHS1-3) in the C-MAF 5′-upstream region (Figure 1C) (30). DHS1 encompasses a broad region containing three peaks located between 78,189,000 and 78,194,500 nucleotides (−2388bp upstream of TSS). DHS2 is between 78,232,000 and 78,233,000 (−39,888bp upstream of TSS) and DHS3 is between 78,361,500 and 78,362,500 (−169,388bp upstream of TSS). Analysis of the sequences within and around the three DHS sites revealed the presence of 5 consensus GAS motifs. GAS1 and GAS5 were within the DHS1-3 peaks whereas GAS2, GAS3, and GAS4 were nearby (Figure 1C). Analysis of the sequence of these GAS motifs revealed strong evolutionary conservation of GAS1 and GAS2, suggesting that these may be important regulatory sequences, while GAS3-5 was less well conserved (Figure S1).

To understand the if these GAS motifs coincided with transcriptionally active chromatin during T cell differentiation, we undertook ChIP-seq analysis of fully differentiated TH1 and TH2 cells using an anti-H3K4me3 antibody and compared sites of enrichment with previously published genome-wide CD4 ChIP-seq data for H3K4me1, H3K4me3 and H2AZ enrichment (10). These studies revealed that GAS1, 2, 4 and 5 were located within regions enriched for monomethylated histone H3K4 while GAS 1, 2 and 4 were also coincident with regions enriched for H3K4me3 in both TH1 and TH2 cells (Figure 1C). All of the GAS motifs coincided with regions enriched for binding of the histone H2 variant H2A.Z. Thus, despite being located a long distance upstream of the C-MAF TSS, the distal GAS motifs mapped to regions containing chromatin modifications consistent with potential enhancer function.

The TSS (left) and the 5′-upstream region of the c-maf gene is depicted with the aforementioned sequences and positions of DHS-associated GAS motifs (GAS1-5), and a negative control GAS6, which is a random, consensus GAS motif (Figure 2A). To determine if IL-2 induces in vivo binding of STAT5 to each of the GAS motifs, we performed a ChIP analysis with anti-STAT5a or control antibody on unstimulated or IL-2-stimulated, PHA-activated human CD4 T cells. The immunoprecipitated chromatin was analyzed for the specific enrichment of GAS1-6 motifs by quantitative real-time PCR analysis (qRT-PCR). STAT5a bound to GAS 1-4 in a specific IL-2-inducible manner, but not to GAS5 or the negative control GAS6 (Fig. 2B). The non-specific IgG control did not enrich chromatin with GAS1-6 sequences in IL-2-stimulated cells by ChIP (Fig. 2B). Thus, IL-2 induced STAT5 binding to multiple sites in the C-MAF promoter at varying distances 5′ to the TSS, and these sites coincided with markers of nucleosomal instability and potential enhancer function. We consistently noted that the lowest level of STAT5 binding was to GAS3, suggesting that this GAS motif alone is insufficient for optimum STAT5 binding. Comparison of the GAS3 sequence (TTCTTTGAA) with the optimal STAT5 binding site (TTC(T/C)N(G/A)GAA) reveals an imperfect match and may suggest a need for cooperative tetrameric STAT5 binding to this region. In this regard, we noted that there is an imperfect GAS motif spaced 7 nucleotides away from GAS3 (indicated in italics in Figure 1A), which may facilitate tetrameric STAT5 binding (28, 31).

To elucidate the STAT5-dependent enhancer activity of these GAS motifs, we assessed the ability of transfected STAT5b to activate constructs containing GAS motifs 1-4 in luciferase reporter assays performed in an IL-2R reconstitution system in HEK293T cells (28). These studies revealed that GAS1, GAS2, and GAS4-driven luciferase reporter constructs increased IL-2-induced, STAT5b-dependent transcriptional activation (Figure 2C). A similar pattern of transactivation of these reporter constructs was observed with STAT5a (data not shown). The GAS3-driven luciferase reporter construct was not transactivated by STAT5b, consistent with poor binding observed at this GAS motif (Figure 2B). Thus, IL-2 induced specific STAT5 binding and transcriptional activation from promoter proximal (GAS1, 2) and distal (GAS 4) GAS motifs within the C-MAF 5′-upstream region.

IL-2 specifically induces C-MAF gene expression

The role of IL-2 in C-MAF gene expression was elucidated by comparing C-MAF RNA levels from freshly isolated and PHA-treated CD4 T cells that were untreated or treated with IL-2 for 0.5-12 hrs. C-MAF mRNA expression in these samples was analyzed by qPCR. IL-2 significantly induced C-MAF mRNA expression in both freshly isolated (Figure 3A) and PHA-activated CD4 T cells (Figure 3B) between 2 to 6 hours after stimulation, and as early as 30 minutes after stimulation. To determine if IL-2 induction of C-MAF gene transcription was associated with a concomitant increase in protein expression, we prepared nuclear extracts from pre-activated CD4 T cells that were stimulated with IL-2 for the indicated times and examined levels of C-MAF, phosphorylated STAT5 (pYSTAT5), total STAT5, and the loading control Lamin-B1 by immunoblot analysis. As shown in Figure 3C, C-MAF protein expression was clearly increased by 6 hours after IL-2 treatment and followed the peak increase in tyrosine phosphorylated STAT5 protein in the nucleus at 30 minutes. Of note, we observed the sustained presence of IL-2-activated STAT5 in the nucleus for the duration of the experiment (24 hours) (Figure 3C). Taken together, C-MAF is induced without TCR stimulation by IL-2 alone in freshly isolated, as well as PHA-activated, CD4 T cells.

As the cloned sequence from the STAT5-ChIP maps to a 5′ region distal to the TSS of the C-MAF gene, we wished to determine if IL-2-induced STAT5 may regulate either of the flanking genes, namely WWOX (approximately 380Kb 3′ of C-MAF) or DNCL2B (approximately 940 Kb 5′ of C-MAF). We therefore performed qPCR analysis to examine the IL-2-induced expression of these two genes in CD4 T cells. Neither WWOX nor DNCL2B were induced by IL-2 stimulation (Figure 3D). Thus, IL-2 induced binding of STAT5 to upstream sequences specifically activates the C-MAF gene.

TCR+CD28-mediated activation of C-MAF transcription is dependent on IL-2R signaling

In the above studies, exogenous IL-2 alone efficiently stimulated C-MAF expression in the absence of TCR stimulation. Thus, we next examined whether TCR activation of C-MAF was dependent on IL-2 signaling by using clinically relevant inhibitors of the IL-2 receptor (HAT) or signaling molecule JAK3.

HAT (daclizumab) binds to the α-chain (CD25) of the high affinity IL-2R, inhibiting binding of IL-2 to the receptor complex and causing efficient blockade of JAK3/STAT5 activation (15). CD4 T cells were isolated from buffy coats and stimulated with αCD3+αCD28 in the presence or absence of HAT for the indicated times. As shown in Figure 4A, C-MAF expression is significantly induced at 17 hours after stimulation and HAT blocked this induction. Thus, the αCD3+αCD28 stimulated early activation of C-MAF is dependent on IL-2/IL-2R signaling.

Because IL-2 induction of C-MAF during T cell activation may specifically involve activation of the JAK3/STAT5 pathway or may utilize additional IL-2 dependent pathways, the effects of a potent JAK3 inhibitor R333 (32) during TCR activation was examined. Fresh CD4 T cells were purified and stimulated with αCD3+αCD28 in the presence or absence of R333; and total cell extracts and RNA were prepared for further analysis. The effect of R333 on activation of STAT5 was investigated by western blot analysis of total cell extracts using phospho-STAT5 and pan-STAT5 antibodies. STAT5 activation (pYSTAT5) was detectable as early as 4 hours, increased by 17 hours, and was blocked by R333 (Figure 4B). Thus, the JAK3-specific inhibitor R333 efficiently inhibited STAT5 activation.

RNA samples were analyzed by real-time PCR for C-MAF, IL-2 and CD25 expression. As noted in Figure 4A, αCD3+αCD28 activation of freshly isolated CD4 T cells increased C-MAF transcription at 17 hours after stimulation. JAK3 inhibitor R333 blocked the induction of C-MAF transcription (Figure 4C). The increase in CD25 and IL-2 expression at 17 hours post-treatment, following peak IL-2 expression at 6-8 hours (data not shown) was also significantly reduced by R333 (Figure 4C). Taken together, these data indicated that the induction of C-MAF transcription during T cell activation was dependent on IL-2 signaling and JAK3/STAT5 activation in CD4 T cells.

IL-2 and IL-6 functionally synergize to potently induce C-MAF gene expression

As both IL-2 and IL-6 can independently activate C-MAF expression and subsequently promote TH2 differentiation, we examined if these cytokines functionally cooperate in the regulation of C-MAF. Purified CD4 T cells were stimulated as above with αCD3+αCD28 or in the co-presence of IL-2 and/or IL-6. As expected, C-MAF transcription was induced by αCD3+αCD28 stimulation, and addition of either cytokine alone modestly increased this induction. However, the presence of both IL-2 and IL-6 significantly increased C-MAF transcription in a synergistic manner (Figure 5A). The presence of an inhibitory antibody to IL-2 or CD25 or the JAK3 inhibitor R333 blocked the synergy, indicating the increase was dependent on IL-2/IL-2R signaling.

To understand if this synergistic effect extended to induction of GATA3, the signature transcription factor of TH2 cells, we quantified GATA3 message in purified CD4 T cells stimulated as above with αCD3+αCD28 or in the co-presence of IL-2 and/or IL-6. In contrast to C-MAF, we did not observe any further increase in αCD3+αCD28 induced GATA3 mRNA expression by IL-2 or the combination of IL-2+IL-6 (Figure 5A). Thus, C-MAF transcription is specifically activated by IL-2 signaling during early CD4 T cell activation.

IL-4 expression is dependent on IL-2 signalling during T cell activation

The key TH2 cytokine gene IL-4 is directly regulated by c-maf (19). We wished to clarify the role of IL-2 signaling in IL-4 activation during T cell activation. CD4 T cells were stimulated with αCD3+αCD28 and IL-2 and IL-6 in the presence or absence of the inhibitors anti-IL-2, HAT, or R333. As expected, IL-4 expression is potently induced by the activating stimuli (Figure 5B). However, the addition of anti-IL-2, or HAT, or R333 significantly reduced IL-4 gene expression, indicating that IL-2R signaling is important for activation of IL-4 transcription during T cell activation (Figure 5B).

IL-2 signalling is required for IL-4 production during TH2 cell differentiation

To determine whether IL-2 signaling was required for IL-4 protein expression during in vitro TH2 cell polarization, we purified naïve CD4 T cells from cord blood samples from three donors and performed in vitro differentiation experiments under TH2 polarizing conditions or TCR (αCD3+αCD28) +IL-2 stimulation, in the presence or absence of the IL-2R-blocking HAT antibody. Fourteen days after treatment, cells were analyzed for cytokine production by flow cytometry. A representative plot is depicted in Figure 6A and the pooled results from all three donors are presented in Figure 6B. As expected, αCD3+αCD28 activation of naïve T cells induced the production of both key TH1 (IFN-γ) and TH2 cytokines (IL-4) (Figure 6A) (26). HAT treatment significantly reduced the percentage of IFN-γ and IL-4 expressing cells, consistent with previous studies using PBMC (15). After 14 days under TH2 conditions, CD4 T cells produced predominantly IL-4, not IFN-γ (Figure 6A). In the presence of HAT, IL-4 was significantly inhibited by approximately 90% in all three donors (Figure 6A and B). Furthermore, if exogenous IL-2 was omitted in the skewing experiments, the cells were unable to polarize to IL-4-producing cells (supplementary Figure S2). Thus, these results indicated that IL-2 signaling plays an important role in the in vitro TH2 differentiation of naïve CD4 T cells.