Dynamic demethylation of the IL2RA promoter during in vitro CD4+ T cell activation in association with IL2RA expression

ABSTRACT

IL2RA, a subunit of the high affinity receptor for interleukin-2 (IL2), plays a crucial role in immune homeostasis. Notably, IL2RA expression is induced in CD4+ T cells in response to various stimuli and is constitutive in regulatory T cells (Tregs). We selected for our study 18 CpGs located within cognate regulatory regions of the IL2RA locus and characterized their methylation in naive, regulatory, and memory CD4+ T cells. We found that 5/18 CpGs (notably CpG + 3502) show dynamic, active demethylation during the in vitro activation of naive CD4+ T cells. Demethylation of these CpGs correlates with appearance of IL2RA protein at the cell surface. We found no influence of cis located SNP alleles upon CpG methylation. Treg cells show constitutive demethylation at all studied CpGs. Methylation of 9/18 CpGs, including CpG +3502, decreases with age. Our data thus identify CpG +3502 and a few other CpGs at the IL2RA locus as coordinated epigenetic regulators of IL2RA expression in CD4+ T cells. This may contribute to unravel how the IL2RA locus can be involved in immune physiology and pathology.

Introduction

IL2RA and its ligand, IL2, are essential to lymphocyte differentiation and immune homeostasis [1,2]. While IL2RA is not expressed in resting T cells, transcription is strongly induced in activated CD4+ T cells by antigenic or mitogenic stimuli [3]. After T cell activation, the binding of IL2 to IL2RA enhances and prolongs IL2RA expression. This allows high concentrations of IL2 to participate in Fas-mediated activation-induced cell death [4]. IL2RA is constitutively expressed in Treg CD4+ T cells (FOXP3+ CD25+) [5].

Six specific regions called positive regulatory regions (PRR), located in cis, mediate the effect of TCR activation on IL2RA transcription through the activation of IL2, TGFβ͵ IFNγ expression [3,6]. The IL2RA gene is a susceptibility gene locus for type 1 diabetes (T1D) [7], rheumatoid arthritis [8], and multiple sclerosis [9], but the mechanisms linking the associated SNPs with changes in gene expression remain unknown.

The first wave of IL2RA transcription in naive CD4+ T cells is triggered by a TCR-CD3 engagement and co-stimulatory signals mainly induced by CD28. This early ‘T cell priming’ activates transcription factors (NF-kB, AP-1, NFAT) [3], which increase the expression of cell surface receptors, including IL2RA, and of multiple cytokines, including IL2 [10]. In turn, IL2 activates IL2RA transcription.

A soluble form of IL2RA (sCD25) increases during T cell activation in vitro [11]. Elevated levels of sCD25 can be observed in the serum of patients having autoimmune diseases such as type 1 diabetes or multiple sclerosis [12,13].

Developmental changes in CpG methylation are known to be a major driver of cell plasticity [14,15]. Indeed CpG methylation, notably CpG-poor regions of the genome have cis-regulatory effects on multiple gene transcription [16,17]. Variation in specific CpG methylation occurs in promoter or enhancer regions in response to genomic variation or changes in cell environment [14,15,18]. We tested whether the methylation of the CpG-poor IL2RA regulatory region has a regulatory role on IL2RA expression. We focused on the 18 CpGs located within the proximal PRR and enhancer regions of the IL2RA gene locus [3]. We characterized the methylation pattern of these CpGs in naive, memory, and regulatory CD4+ T cells, then studied the effects of in vitro activation on naive CD4+ T cell methylation and the correlation of methylation with IL2RA expression.

Materials and methods

Subjects

Isis-Diab is a prospective cohort launched by the Programme Hospitalier de Recherche Clinique of the French Ministry of Health in 2008 with the objective of studying genetic, epigenetic, and environmental risk factors of childhood-onset autoimmune T1D. Thirty-six T1D patients and 40 controls of Caucasian ancestry were included in our study. Two groups of control subjects were recruited: 11 young controls age-matched to T1D patients, and 29 older controls allowing to study the effects of age on CpG methylation (Figure 1(b)). In addition, seven control subjects (mean age 18.8 ± 11 years) were used to study CpG methylation in various immune cell types. In T1D subjects, the mean age at diagnosis was 8.0 ± 3.9 years, a mean diabetes duration of 3.8 ± 4.5 years and a mean hemoglobin A1c (HbA1c) of 8.5 ± 1.8% at time of sampling. The research protocol was approved by the Ethics Committee of Ile de France (DC-2008–693) and the computer security and confidentiality guarantees given to patients was approved by the Commission Nationale Informatique et Libertés (DR-2010–0035). The clinicalTrial.gov identifier was NCT02212522. All patients provided written informed consent for participation in the study and donation of samples. We obtained written informed consent from the next of kin, caretakers, or guardians on behalf of the children enrolled in the study.

Figure 1.

(A) Levels of methylation of the 18 studied CpGs in three subpopulations of CD4+ T lymphocytes: naive CD4+ T cells (n = 76, black dots), memory T cells (n = 5, blue dots) and T regulatory cells (n = 7, red dots). The x-axis shows the studied CpGs (black lollipops), the positive regulatory regions (PRR), the negative regulatory region (NRE), and the TSS (black arrow). (B) Table showing the comparison of methylation levels for the 18 CpGs in the studied groups of T1D patients and controls.

Cell isolation

Venous blood samples were collected and peripheral blood mononuclear cells (PBMCs) were immediately purified from fresh blood (12–25 mL) for naive and memory CD4+ T cells or 50 mL for Tregs isolation using lymphoprep density gradient (Lymphocytes separation medium, Eurobio, CMSMSL01-01).

Untouched naive CD4+ T cells were purified by magnetic indirect isolation using the Naive CD4+ T cell Isolation Kit II (Miltenyi, 130–094-131). Isolation process involves a depletion of non-CD4 T cells, memory CD4+ T cells, and CD25+ cells. These cells are indirectly magnetically labeled with a cocktail of biotin-conjugated monoclonal antibodies before anti-Biotin MicroBeads are added. The magnetically labeled non-naive CD4+ T cells are depleted by retaining on a magnetic column while the unlabeled naive CD4 + T cells pass through the column. We used anti-human CD4-FITC, CD45RA-PE, CD197-APC (all from eBioscience, 11–0049, 12–0458, 17–1979) to characterize the naive CD4+ T cell population (CD4+ CD45RA+ CCR7+). Purity of naive CD4+ was > 92% using flow cytometry (Accuri™ C6, BD Biosciences) (Fig. S1A).

Human CD4+ CD25+ T cells were isolated from 50 mL of peripheral blood of 7 blood donors using density gradient centrifugation and purified by Human CD4+ CD25+ regulatory T Cell Isolation Kit (Miltenyi 130–091-301) in a two-step procedure. First, non-CD4+ T cells were depleted from the total PBMCs. Then, CD25+ cells were labeled with anti-CD25 magnetic microbeads to select CD4+ CD25+ T cells. Positive cells were applied to a second magnetic column, washed, and eluted again. This procedure achieved a selective purification of CD4+ CD25+ T cells (Tregs).

Human CD4+ memory T cells were isolated from 25 mL of peripheral blood of 5 blood donors using density gradient centrifugation and purified by magnetic indirect isolation using the CD4+ Central Memory T cell Isolation Kit (Miltenyi 130–094-302). Briefly, memory CD4+ T cells were collected after non-CD4+ T cells and naive CD4+ T cells depletion.

Five major blood cell populations were isolated from 7 blood donors: CD4+ T cells, CD8+ T cells, CD56+ Natural Killer cells, CD19+ B lymphocyte cells and CD14+ monocytes. For this purpose, PBMCs from 15 mL of blood were separated in 2 parts used for a positive magnetic selection of CD19+ and CD56+ cells with microbeads. The magnetically labeled cells were retained on the column and eluted as the positively selected cell fraction. The unlabeled sorting cells were pooled then separated in 3 parts and used for a second positive selection with CD14+, CD8+, or CD4+ microbeads according to the same process. All reagents were from Miltenyi (Paris, France) (130–045-101, 130–045-201, 130–050-201, 130–050-301, 130–050-401).

Activation of naive CD4+ T cells

Naive CD4+ T cells were cultured (1 x 10⁶ cells/well in a 24 well plate) in RPMI medium supplemented with 10% FCS, penicillin/streptomycin, Hepes, β-mercaptoethanol, and stimulated using anti-CD3/anti-CD28-coated beads (Dynal, Invitrogen, 111.31D) according to the manufacturer’s instructions (1: 1 cell bead ratio). Control samples were not stimulated. Cells were incubated at 37°C for 24 hours. Cells were harvested and aliquots were taken for activation analysis by flow cytometry or pelleted for DNA extraction and supernatants were frozen after clarification by centrifugation.

In another experiment, we performed a more detailed time course study of activation of naive CD4+ T cells at 0, 2, 4, 6, 12, 15, 18, 21, and 24 hours following in vitro activation.

Choice of CpGs

IL2RA locus contains no CpG island and can be described as a ‘CpG poor’ or ‘CpG low’ region [19,20]. We selected 18 CpGs located within or close to the regulatory regions of the IL2RA locus: 6 CpGs located −241 to −459 in PRRI-PRRII (2 CpGs) or NRE (4 CpGs) [21], 5 CpGs around PRRIII [22,23], 2 CpGs at PRRV [24], 4 CpGs at PRRVI (CD28re) around −8500 bp [25], and 1 CpG (+ 3502) in PRRIV [26]. See Figure 1(a) and Table S1 for precise positions of the CpGs.

DNA methylation

DNA from cells was isolated by alcohol precipitation after cell lysis using Gentra Puregene Cell Kit (Qiagen, 158445) following manufacturer’s instructions. DNA concentration was determined by spectrophotometry on the NanoDrop 2000. Genomic DNA (200 ng) was treated with EZ-96 DNA Methylation-Gold Kit, according to manufacturer’s protocol (Zymo Research Corporation, D5005).

We PCR-amplified the bisulfite treated genomic DNA using unbiased primers (Tables S1 and S2) and performed quantitative pyrosequencing: biotin-labeled single stranded amplicons were isolated according to protocol using the Qiagen Pyromark Q96 Work Station and underwent pyrosequencing with 0.5 µM primer. Pyrosequencing was performed using a PyroMark Q96 ID Pyrosequencing instrument (Qiagen). The percent methylation for each of the CpGs within the target sequence was calculated using PyroQ CpG Software (Qiagen, Germany).

IL2RA expression

On a part of naive CD4+ T cells (n = 28), we analyzed IL2RA cell surface expression after α-CD3/α-CD28 stimulation or in basal conditions (24 hours) by flow cytometry with anti-human CD25-PE (Miltenyi, 130–098-211) labeling.

IL2 and soluble IL2RA immunoassays

Cell culture supernatants were collected from unstimulated and stimulated-naive CD4+ T cells. IL2 and soluble IL2RA concentrations were measured by enzyme-linked immunoabsorbent analysis using the Quantikine immunoassay kits (R&D Systems, D2050, DR2A00) according to manufacturer’s instructions.

Genotyping

We studied the 7 SNPs associated with type 1 diabetes [27] and other autoimmune diseases [28–30]. Two of these 7 SNPs (rs11594656 and rs41295061) are located in an intergenic region between IL2RA and RBM17; the others are located in the first intron.

The rs11594656, rs41295061, rs3118470, rs3134883, rs2104286, and rs706778 SNPs were genotyped by TaqMan® pre-designed SNP genotyping assay technology under conditions recommended by the manufacturer (LifeTech, 4351379). The rs12722495 was genotyped using the pyrosequencing method. Primers for SNP region amplification are: F-GGACTCTCTGGGGAACAGAAC and R-CTGCTTGAGCATACCTGTTCC (with F-primer biotinylated). PCR is performed with recombinant Taq DNA Polymerase (Thermo Fisher Scientific, EP0402) and (NH₄)₂SO₄ buffer, using a temperature of 60°C for the annealing step. Biotin-labeled single stranded amplicon (160 bp) was isolated according to manufacturer’s protocol using the Qiagen Pyromark Q96 Work Station and underwent pyrosequencing with 0.5 µM primer (5ʹ-CTCTAATTTGGTGAGTTTCA-3ʹ). Analyses were performed with the accompanying SNP software.

Statistics

Demethylation was normalized to basal methylation in naive CD4+ T cells. The relative decrease in methylation is calculated as follow: methylation before stimulation – methylation following stimulation divided by methylation before stimulation.

CpG methylation was compared across IL2RA SNP alleles using non-parametric Kruskal-Wallis test. Student’s t-test was used to compare groups when data were in sufficient numbers and normally distributed.

Wilcoxon Rank Sum test was used to test the null hypothesis between two different groups when sample size was small.

All statistical analyses were conducted using R 2.14.2.

Results

Basal CpG methylation at the IL2RA locus in T1D and controls

The table in Figure 1(b) shows that CpG methylation in naive CD4+ T cells is comparable in T1D and controls at all studied positions. Diabetes duration and HbA1c showed no association with CpG methylation. T1D and control groups were thus pooled for further analyses of CpG methylation in naive CD4+ T cells. The methylation status of the 18 studied CpGs in the three studied T lymphocyte subpopulations is presented in Figure 1(a) and Table 1.

Table 1.

CpG methylation at the IL2RA promoter in naive, Treg, and memory CD4+ T cells.

	CD4+ T cells
CpG site	Naive (n = 76)	Memory (n = 5)	Regulatory (n = 7)	P value
CpG +3502	21.8 ± 5.4	8.9 ± 1.1	4.1 ± 1.1	2.10−4(a) 3.10−6(b) 4.10−3(c)
CpG −241	1.9 ± 1.7	1.5 ± 0.5	7.6 ± 2.5	0.57(a) 6.10−6(b) 1.10−3(c)
CpG −272	1 ± 1.6	2.1 ± 0.6	0.7 ± 1.2	0.017(a) 0.56(b) 0.07(c)
CpG −356	4.2 ± 1.7	2.8 ± 0.5	1.7 ± 1.1	0.016(a) 1.10−4(b) 0.027(c)
CpG −373	35.2 ± 4.5	10.1 ± 1.6	10.9 ± 2.5	2.10−4(a) 4.10−6(b) 0.72(c)
CpG −456	93.7 ± 2.7	48.8 ± 3.3	45.8 ± 4.5	3.10−4(a) 7.10−6(b) 0.3(c)
CpG −459	85.4 ± 2.5	62.2 ± 3.3	51.8 ± 10.3	3.10−4(a) 7.10−6(b) 0.01(c)
CpG −3691	78.6 ± 8	28.7 ± 3.6	15.2 ± 1.4	2.10−4(a) 5.10−5(b) 4.10−3(c)
CpG −3789	91.5 ± 7.1	41.6 ± 5.3	17.8 ± 4.1	2.10−4(a) 5.10−5(b) 4.10−3(c)
CpG −3851	83.3 ± 8	39.2 ± 4	18.6 ± 4.4	2.10−4(a) 5.10−5(b) 4.10−3(c)
CpG −3946	74.9 ± 4.6	48.4 ± 3	32.4 ± 4.2	2.10−4(a) 6.10−5(b) 4.10−3(c)
CpG −3970	91.1 ± 6.6	48 ± 4	32.7 ± 4.6	2.10−4(a) 5.10−5(b) 4.10−3(c)
CpG −7494	88.4 ± 2	81.8 ± 2.3	65.8 ± 5.5	2.10−4(a) 5.10−5(b) 4.10−3(c)
CpG −7591	67.7 ± 4.3	67.2 ± 2	34.3 ± 5.1	0.8(a) 5.10−5(b) 4.10−3(c)
CpG −8452	59.2 ± 4	54.4 ± 3.6	27.7 ± 4.1	0.01(a) 5.10−5(b) 4.10−3(c)
CpG −8476	87 ± 4.5	79.5 ± 3.1	45.1 ± 4.8	2.10−3(a) 5.10−5(b) 4.10−3(c)
CpG −8482	61.7 ± 4.8	47.2 ± 4.4	24.3 ± 3.1	3.10−4(a) 5.10−5(b) 4.10−3(c)
CpG −8564	33.2 ± 3.6	31.5 ± 2.4	16.9 ± 2.2	0.2(a) 5.10−5(b) 4.10−3(c)

Data are expressed as mean ± SD. P values were calculated with Wilcoxon rank test. (a) = naive vs. memory CD4+ T cells; (b) = naive vs. regulatory T cells; (c) = memory vs. regulatory T cells.

The 3 CpGs closest to the transcription start site (−241, −272, −356) were largely unmethylated, with a methylation level ranging 1–8% in all studied T cell populations. The other 15 CpGs had variable levels of methylation in naive CD4+ T cells, memory cells, and Treg cells.

In naive CD4+ T cells, we found that 9/18 CpGs (CpGs −8476, −7494, −3970, −3946, −3851, −3789, −3691, −459, −456) were highly methylated (75–94%), while 6/18 CpGs (−8564, −8482, −8452, −7591, −373, + 3502) had intermediate levels of methylation (22–68%).

In memory CD4+ T cells, the level of CpG methylation was slightly lower but comparable to that measured in naive CD4+ T cells for PRRVI CpGs (−8564, −8482, −8476, −8452) and PRRV (−7591, −7494). For the other CpGs, methylation levels of memory T cells were comparable to those in Treg cells (Table 1).

Treg cells showed a much lower methylation level in almost all CpGs compared with naive CD4+ T cells (3.10⁻⁶< P <1.10⁻⁴), with the largest difference for the 5 CpGs located around PRRIII (−3970, −3946, −3851, −3789, −3691).

Methylation showed a wide inter-individual variation at CpGs −3970, −3851, −3789, −3691, and + 3502, with a coefficient of variation (standard deviation/mean) ranging 7–24% (data not shown). These CpGs can thus be considered ‘inter-individual DMRs’ [31]. In contrast, other CpGs showed little inter-individual variation.

Methylation was closely correlated across CpGs located within PRRIII and PRRVI. Methylation of CpGs at PRRIII CpGs also correlated with methylation at CpGs −241 and −456 (Fig. S2).

The CpG methylation pattern in other immune cell types is presented in Table S3. Briefly, in CD4+, CD8+ T cells, natural killer cells (NK), B cells, monocytes, and macrophages, studied in only 7 individuals, no significant difference in CpG methylation was observed in PRRI and PRRII. CpG +3502 methylation was 1.5-fold higher in NK cells (P = 0.05), 2-fold higher in B cells (P = 0.003), and 3.3-fold higher in monocytes (P = 0.005) than in CD4+ T cells. The methylation pattern varied across monocytes, B cells, and CD4+ T cell populations (P = 1.10⁻⁵ and P = 3.10⁻⁴, respectively) at PRRVI. Large differences between NK cells, monocytes and CD4+ T cells (P = 2.10⁻¹² and P = 6.5.10⁻¹³, respectively) were observed at PRRIII. CD4+ and CD8 + T cells showed comparable methylation levels at all studied CpGs.

We found no statistically significant correlation between the methylation status of CpGs in naive CD4+ T cells and alleles of any of the 7 SNPs known to be associated with T1D or other autoimmune diseases (Figure 2, Fig. S3 and LD matrix in Fig. S4).

Figure 2.

Lack of correlation between CpG +3502 methylation and genotypes for the 7 studied SNPs at the IL2RA locus. The asterisk (*) indicates the two closest SNPs flanking CpG +3502.

In vitro activation of naive CD4+ T cells

Following in vitro stimulation of naive CD4+ T cells, methylation undergoes a large decrease at 5 CpGs sites: the 4 CpGs of PRRVI (−8564, −8482, −8476, −8452) (Fig. S5) and the +3502 CpG. Absolute decrease of methylation is 8–20% and relative decrease 13–48% (Figure 3 and Table S4). Most of the demethylation takes place between 6 and 18 hours of activation (Figure 4). In contrast, methylation did not change at all other CpG sites (Fig. S6 and Table S4). T1D and controls showed a similar pattern and degree of CpG demethylation during activation of their naive CD4+ T cells: demethylation at CpG +3502 was 48 ± 1% in T1D, 47 ± 1% in young and older controls; demethylation at PRRVI CpGs was 20 ± 1% in T1D, 21 ± 1% in young controls and 16 ± 1% in older controls.

Figure 3.

The four PRRVI CpGs (−8452 to −8564 bp) and the PRRIV CpG +3502 show significant demethylation (P <3.10⁻¹⁶) following in vitro stimulation of naive CD4+ T cells (T1D ○, n = 36; children controls ●, n = 8; Adult controls ■, n = 32).

Figure 4.

Time course of CpG +3502 demethylation during in vitro activation of naive CD4+ T cells. Each point represents the mean ± SD of methylation at CpG +3502 in 3 independent in vitro experiments.

We found that IL2RA protein is not detected on naive CD4+ T cells and was expectedly expressed after activation for 24 hours. The proportion of IL2RA positive cells after activation showed a wide variation across individuals, ranging 7–89% of gated CD4+ CD45RA+ CCR7 + T cells determined by flow cytometry (Fig. S1B). The proportion of IL2RA-positive cells correlated with the decrease in CpG +3502 methylation (R = 0.52, P = 3.10⁻³) (Figure 5(a)). A correlation between IL2RA surface expression with the decrease of methylation was also observed for the CpGs located within PRRVI (−8564, −8482, −8476, −8452).

Figure 5.

The demethylation occurring at CpG +3502 correlates with (a) CD4+ T cell expression of IL2RA at the cell surface of CD4+ T cells activated in vitro (R = 0.52, P = 3.10⁻³, n = 28) and (b) the concentration of IL2RA in cell supernatant (R = 0.51, P = 4.10⁻⁵, n = 52).

The concentration of soluble interleukin-2 receptor (sIL2RA) in supernatant during activation of naive CD4+ T cells was correlated with CpG +3502 demethylation (R = 0.51, P = 4.10⁻⁵) (Figure 5(b)).

We also observed a significant correlation between IL2 concentration in activated-CD4+ T cell supernatant and CpG +3502 demethylation (R = 0.46, P = 3.10⁻⁴) (Figure 6).

Figure 6.

The decrease in CpG +3502 methylation correlates with the concentration of IL2 in supernatant of CD4+ T cells activated in vitro (R = 0.46, P = 3.10⁻⁴, n = 59).

In response to activation, the decline in methylation at CpG +3502 correlated with a parallel decrease of methylation at the CpG −252 located in the IL2 promoter (Figure 7).

Figure 7.

Parallel time course of the demethylation of CpG +3502 at the IL2RA promoter (solid line) and demethylation of CpG −252 at the IL2 promoter (dotted line).

Effect of age

Age was associated with basal methylation at several CpGs of the IL2RA promoter. Notably, CpG +3502 methylation decreased by 47% from early childhood (<10 years of age) to late adulthood (≥38 years of age) in naive CD4+ T cells (R = 0.47, P = 3.10⁻⁵). The other 8 CpGs showing significant age-dependent decrease of methylation are presented in Figure S7. A decrease of methylation with age was not observed in other cell types, in a limited sample of individuals (n = 7), which limits the robustness of this observation.

Age showed no significant effect on the degree of demethylation occurring in vitro during the activation of CD4+ T cells.

Discussion

Our main finding is that specific CpGs located within the IL2RA promoter undergo a strong demethylation when naive CD4+ T lymphocytes are activated in vitro. We were not able to investigate the mechanisms leading to in vitro activation to induce such prominent changes in CpG methylation at CpG +3502 and at CpGs of PRRVI, nor in other CpGs. A second finding is that demethylation at these CpGs was associated with IL2RA expression. A third finding is that age was associated with a decrease in methylation at 9 CpGs of the locus, notably CpG +3502.

We observed that methylation for several CpGs of the IL2RA locus varied across individuals. This variability was observed herein in a homogeneous population of naive CD4+ T cells, but not in a heterogeneous cell mixture [32]. The most variable of all CpGs was CpG +3502 with an inter-individual coefficient of variation of ~20%, not explained by age. This CpG can thus be considered an ‘inter-individual DMR’ [31] able to generate differences in IL2RA expression in CD4+ T cells across individuals.

Our observation is not the first example of the different behavior of CpGs located within the same locus involved in immune responses. Indeed, demethylation of specific CpGs located within CpG islands, CpG-poor, or CpG-rich gene promoters occurred in response to activation at the IL2 locus in CD4+ T cells [18], at the PDCD1 locus in CD8+ T cells [33], at RORC locus in CD45RA- memory Tregs [34], at the CCR6 locus in CD4+ and CD8+ T cells [35], at the IFNγ locus in CD8+ T cells [36], and within the IL4/IL13/RAD50/IL5 gene cluster in Th2 T cells [37,38]. The association of demethylation with gene expression has been reported for several immune genes in mouse and humans. In mouse, the demethylation of CpGs in IL2 promoter enhanced gene transcription [39]. In humans, the demethylation of CpG −252 located in the promoter of IL2 controlled IL2 expression in CD4+ T cells following in vitro activation [18]. Inhibitors of CpG methylation increased IL2 gene expression [40]. CpG hypomethylation or demethylation in the promoter of IFNγ correlated with IFNγ expression in memory and CD8+ T cells, respectively [41,42]. Changes in CpG methylation occurred during lymphocyte polarization [43,44].

CpG demethylation can be a passive or an active process. Passive CpG demethylation occurs throughout DNA replication and is due to lack of methylation maintenance during cell cycles.

Active CpG demethylation is key to major developmental and physiological processes as genome stability and defense, imprinting, and X chromosome inactivation [15,45]. The current results show that the observed dynamic demethylation is active. Active demethylation involves Tet2, which induces CpG demethylation in Th specific region and regulates cytokine expression during Th cell differentiation [46].

The relationship between CpG demethylation and activation of transcription involves changes in chromatin structure and recruitment of transcription factors. At the FOXP3 locus, the recruitment of Stat5 and CREB at cognate unmethylated DNA regions allows FOXP3 transcription in Treg cells. In non-Treg cells, the same regions are highly methylated, which inhibits transcription factor binding and FOXP3 transcription [47].

Recent studies revealed that Tet1 and Tet2 mediate FOXP3 demethylation and drive Treg differentiation [48]. At the IL2 locus, the demethylation of CpG −252 allows the binding of Oct-1 followed by the binding of other transcription factors and activation of IL2 transcription [18]. At the IL2RA locus, the intronic regulatory region containing CpG +3502 binds several transcription factors, notably Stat5b [26,49]. Given the proximity of CpG +3502 15 bp from the STAT5 binding site, it is conceivable, although entirely speculative, that the methylation of CpG + 3502 modifies the binding of STAT5 to its binding site.

In the current study, demethylation at specific CpGs at the IL2RA locus, including CpG +3502, is associated with the activation of transcription and the expression of the IL2RA protein. This observation supports that demethylation of a few CpGs acts as a transcriptional regulatory mechanism controlling IL2RA expression. It is possible though that other unidentified cis located CpGs contribute to this transcriptional regulation. It is noteworthy that demethylation in both IL2RA and IL2 promoters occurred in parallel in CD4+ T cells as if it was controlled by coordinated and synchronous signals.

Circulating memory T cells, which do not express IL2RA in the basal state, and Treg cells, which express IL2RA in the basal state, both showed low levels of methylation at specific CpGs of the IL2RA locus, notably at CpG +3502. We have not studied the response of memory CD4+ T cells to activation, thus do not know whether demethylation at the IL2RA promoter could mediate adaptive immune responses via the rapid production of effector cytokines by these cells.

In human Tregs [50], a much lower CpG methylation at the IL2RA promoter compared with CD4+ T cells or PBMCs has recently been reported [51]. A recent study found that the IL2RA promoter is demethylated in mice Tregs [52]. It is noteworthy that we have not studied Treg cells in T1D patients.

Other immune cells – CD4+ T cells, CD8+ T cells, natural killer cells, B cells, and macrophages – do not express IL2RA in the basal state [1]. The differences in methylation were observed at specific CpGs across these cell types, suggesting that methylation can contribute to the regulation of the response of these cells to IL2.

Since in certain regions of the genome CpG methylation is influenced by cis genetic variation [53], we tested whether this is true at the IL2RA promoter locus. Indeed, genome-wide association studies have revealed that the IL2RA region is a susceptibility locus for several diseases, such as T1D, multiple sclerosis, vitiligo, or rheumatoid arthritis [7,30,54–59], but no one has demonstrated how sequence variants could act on IL2RA gene expression. The only exception is that Schwartz et al. [60] identified a relationship between rs61839660 and transcription factor binding in the first intron of IL2RA gene, where CpG +3502 is located. It is thus conceivable that genetic variation and epigenetic variation across individuals combine their effects at this position without being dependent on each other.

We found that the basal methylation of the IL2RA promoter in naive CD4+ T cells, as well as the activation-mediated demethylation, do not seem to be influenced by the cis variation of the DNA sequence. Thus, it seems unlikely that the individual variation of methylation at CpG +3502 or other CpGs could explain the association of the 7 intergenic and intronic SNPs with autoimmune phenotypes. This conclusion was reached, for example, by comparing CpG +3502 methylation (22 ± 5%) in 32 people carrying the AG/AA genotype, with methylation (21.7 ± 5.8%) in people carrying the GG genotype. It is unlikely that enlarging the size of the studied population will alter our observation. Again, the lack of genetic-epigenetic association has only been investigated herein in naive CD4+ T cells, but not in Treg cells or other immune cells.

Besides genetics, CpG methylation can be modulated by environmental signals [14,15]. In addition, some CpGs gain or lose methylation with time [61]. Indeed, while CpG methylation is maintained over cell divisions, a gradual loss called ‘epigenetic drift’ is observed over the lifespan and varies across individuals [62]. In contrast, ‘epigenetic clock’ is another phenomenon describing the parallel decline of CpG methylation in body tissues in all individuals [61]. In our study, specific CpG sites showed demethylation with increasing age in all individuals, notably CpG +3502 and 3/5 CpGs at PRRIII. In contrast, other CpGs showed no correlation with age. This age-related decline was not observed in the other studied immune cells, suggesting that only naive CD4+ T cells are subjected to age dependent changes at these CpGs. Our interpretation is that these CpGs obey an epigenetic clock that is specific to naive CD4+ T cells that is not shared by other immune cells and is observed in all individuals. More extensive studies of the effects of age would require to enlarge the population of interest to patients of older ages. It is conceivable that the age-dependent decrease in methylation in naive T cells, if confirmed in a larger population, contribute to the propensity of autoimmunity with advancing age [63], by allowing CD4+ T cells to respond to IL2.

In summary, the methylation of specific CpGs of the IL2RA locus appear to sustain important regulatory mechanisms for CD4+ T cell activation, IL2RA expression, and age-related immune homeostasis in humans.

Funding Statement

This work was supported by the Novo Nordisk France [Conv_Fondation_Recherche_APHP];Lilly France [VAL 2014/2014-014/01]; ANR RESET-AID Project (ANR-15-EPIG-0004).

Acknowledgments

This work was supported by institutional grants from NovoNordisk France and Lilly France to the « Fondation de l’AP-HP pour la Recherche » and ANR RESET-AID Consortium Project (REsolving Systems Epigenomes of T-cells from Autoimmune and Inflammatory Disease patients) 2. MPB is supported by the Fondation de l’AP-HP pour la Recherche. We acknowledge C. Auffray and F. Adam for support with flow cytometry and we thank Clémence Mille and Clémence Qu’hen for logistical help in cell sorting.

Authors’ contributions

PB and MPB designed the study, interpreted the data, and wrote the article. MPB performed molecular studies. ALC and SLF recruited the patients and shared discussion of the results. SLF did the statistical analyses.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplementary material can be accessed here.