Intron-1 rs3761548 is related to the defective transcription of Foxp3 in psoriasis through abrogating E47/c-Myb binding

Abstract

Foxp3 is a master transcription factor (TF) for development and function of CD4⁺CD25⁺Foxp3⁺ regulatory T cells (Treg cells) and is critical for the transcription of target genes. But the transcriptional regulation of Foxp3 itself has not been fully understood until now. Here, we aimed to demonstrate the hypothesis that upstream single nucleotide polymorphism(s) (SNPs) of Foxp3 was/were responsible for the defective transcription of Foxp3 in psoriasis and to explore the mechanism behind this hypothesis. In this study, SNP of large sample was investigated for risk analysis. Mature algorithms, electrophoretic mobility shift and chromatin immunoprecipitation assays were used to identify TF binding site variations. Loss-of-function and overexpression assays and cell cycle blocker assay were performed to identify when and what kind of possible roles the candidate factors play. Our results showed that intron-1 rs3761548 was correlated with a significant susceptibility to psoriasis. The rs3761548 contributed to the decreased resting Foxp3 transcription and impaired acceleration of Foxp3 transcription levels after stimulation in psoriatic patients with genotype AA. We analysed and demonstrated potent new E47/c-Myb -dependent regulation elements in rs3761548, oppositely controlling Foxp3 gene transcription at G1 and G2/M phases of Treg cells in psoriatic patients. For patients with rs3761548 AA, the polymorphism causes loss of bindings to the E47 and c-Myb factors, leading to defective transcription of Foxp3 gene. Further identification of the networks and molecular mechanisms underlying Foxp3 transcription may provide new insights into Foxp3 transcriptional regulation and alternative therapeutic strategies to improve characteristics of autoimmune disorders.

Introduction

Regulatory T cells (Treg cells) are a well-characterized subpopulation of T cells that exert immunosuppressive effects to control the immune response. Deficiency or dysfunction of Treg cells can predispose to autoimmune diseases [1]. Psoriasis vulgaris (PV) is a common dermatological disorder, affecting approximately 1–3% of the general population. The lesion appears as patches of raised, reddish and itching skin covered by silvery-white scale, which symmetrically occurs anywhere on the body, even on faces. Living with this lifelong condition can be physically and emotionally challenging. What’s more, awareness is increasing that PV is more than skin deep and that it is associated with other systemic autoimmune disorders [2–4]. Recent compelling evidence suggest that PV is one of the commonest immune-mediated autoinflammatory diseases and activated T cells play an important role in the progression of PV [5, 6]. The balance between regulatory and effector functions is important for maintaining efficient immune responses, but in psoriasis, the imbalance is observed. Our study (data not shown) and a variety of published studies suggested disproportionate and /or dysfunctional blood and target tissue CD4⁺CD25⁺Foxp3⁺ Treg (Tri-Treg) cell activity led to reduced restraint and consequent hyper-proliferation of pathogenic T cells in vivo[7–9].

Foxp3, forkhead/winged helix transcription factor 3, is a master control transcription factor for development and function of Tri-Treg cells and is critical for transcriptional repression [10]. Foxp3 can interact with other important transcription factors (TF) to repress cytokine gene expression [11]. Until now, the regulation of Foxp3 itself is not fully understood. It has been shown that Foxp3 expression can be induced in the CD4⁺CD25^– population by corticosteroids [12], oestrogen [13] and TGF-β (transforming growth factor β) [14], suggesting that Foxp3 can be induced in peripheral T cells. Recently published studies of Foxp3 regulation have revealed a proximal 5′ regulatory region that was reported to contribute to T-cell receptor (TCR)-mediated regulation of the gene [15], whereas interleukin-2 (IL-2)-induced Foxp3 expression has been attributed to intronic tandem signal transducer and activator of transcription (STAT) binding motifs in vitro[16]. A TCR-responsive enhancer containing a CpG island in the Foxp3 first intron also has been identified [17]. These results give us a hint that, besides promoter, the intronic motifs may also play an important role in the regulation of Foxp3 transcription. Due to the accessibility of both peripheral blood and target tissue cells involved in the pathogenic process, PV offers a unique opportunity to study the mechanism how Foxp3 transcription is regulated in human disease.

In this study, we performed single nucleotide polymorphism (SNP) investigation of large sample in PV patients and found a potent regulation site in intron-1 that influenced Foxp3 transcription levels in psoriatic patients with different genotypes. Enlightened from these results, we analysed and demonstrated potent new E47/c-Myb-dependent regulation elements in intron-1. E47 and c-Myb bound to these overlapping elements at G1 and G2/M cycle windows in Treg, respectively, and oppositely control Foxp3 gene transcription levels in PV.

Materials and methods

Agents and antibodies

Human Treg Flow™ Kit (320401) was from BioLegend (San Diego, CA, USA). APL 119R (Altered peptide ligand 119R) [18] was synthesized by Sangon Biotech (Shanghai, China) and >90% pure as determined by high-performance liquid chromatography. Control peptide was GST p61–75 ITDPKKCPPILLYRY, also with >90% purification. HiPerFect transfection reagent was purchased from Qiagen (Qiagen, Hilden, Germany). AIM V® Medium (serum free lymphocyte medium) was from GIBCO (Grand Island, NY, USA). pGL3 basic vector and dual luciferase assay system were the products of Promega (Madison, WI, USA). QuickChange kit for site-directed mutagenesis and ChIP assay kit were from Stratagene (La Jolla, CA, USA) and Upstate (Lake Placid, NY, USA). E2A small interfering RNA (siRNA; sc-35245), c-Myb siRNA (sc-29855) and control siRNA-A (sc-37007) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). The pcDNA_3- vectors containing full-length c-Myb and E2A were kind gifts from Dr. Jia Wu (Zhejiang, China) and Dr. Lin Lin (Shannxi, China). Aphidicolin and Nocodazole were purchased from Calbiochem (San Diego, CA, USA). PE anti-human CD4 (RPA-T4; Biolegend) and FITC anti-human CD25 (BC96; Biolegend) were used for CD4⁺CD25⁺ T cells isolation by flow cytometry. Anti-human Foxp3 (259D; BioLegend), anti-β-actin (10A5; Chemicon), anti-E47, anti-c-Myb, anti-cyclin B1, anti-cyclin D1 and anti-cyclin E antibodies (Yae, H-141, V63, 72–13G and E-4, respectively; Santa Cruz Biotechnology) were for Western blotting. Anti-HEN1, anti-E47, anti-c-Myb and anti-PPAR-γ antibodies (E-13, E-8, Yae and H-141, respectively; Santa Cruz Biotechnology) were for electrophoretic mobility shift assays (EMSAs) and chromatin immunoprecipitation assay (ChIP).

SNP analysis

A total of 524 patients and 549 healthy controls were recruited in SNP study. All the patients had not subjected to UVB and other immunosuppression therapy before. Acquisition of tissues and blood samples was agreed by the patients or their guardians, who had signed the consent form. This study had been approved by the Review Board of Third Military Medical University. The SNPs (rs3761547, rs3761548, rs3761549, rs4824747 and rs5906761) were selected from promoter and intron-1 region to extract the maximum genetic information based on Asian population diversity (minor allele frequency >10%), according to the data from database of International HapMap Project (http://www.hapmap.org/), SNPPER (http://snpper.chip.org/) and NCBI’s Entrez system (http://www.ncbi.nlm.nih.gov/sites/entrez?db=snp). Genomic DNA was isolated from 1 ml anti-coagulated peripheral blood leucocytes using DNA extract kit. Genotyping of these SNP sites was determined using polymerase chain reaction with sequence-specific primers (PCR-SSP) or PCR-restriction fragment length polymorphism (RFLP) methods. The demographic and clinical data were also collected.

Reverse transcriptase-polymerase chain reaction (RT-PCR)

All samples were analysed by standard RT-PCR reactions. The primers used are shown as follows: human Foxp3 forward primer: 5′- ACA CCA CCC ACC ACC GCC ACT -3′ and reverse primer: 5′- TCG GAT GAT GCC ACA GAT GAA GC -3′[19]; human β-actin forward primer: 5′- GCA ATG AGC GGT TCC GCT GC -3′ and reverse primer: 5′- CG ATC CAC GAG TAC TTG -3′. All experiments were conducted three times using the appropriate positive, negative, and no template controls. The cDNA products were verified by agarose gel and DNA sequencing analysis. The ratio between Foxp3 and β-actin were determined to assay the relative levels of Foxp3 gene transcription.

Western blotting

Western blotting was performed as described in related literature [20]. Cell extract was resolved by electrophoresis, transferred to nitrocellulose and probed with primary antibodies, respectively. Proteins were visualized using corresponding secondary antibody conjugated to HRP and a chemiluminescence detection system. Anti-β-actin antibody by the same procedure was used to evaluate the expression levels of Foxp3.

Foxp3 resting and accelerated levels after APL stimulation

By SNP confirmation, there were 11 patients with genotype CC and 3 with AA in the 18 HLA DRB1*07 positive psoriatic patients who had enrolled in the first-part clinical trial. Peripheral blood mononuclear cells (PBMCs) were incubated with or without APL 119R (100 μg/ml) at 37°C for 3 hrs and irradiated (4000 rad) and served as antigen presenting cells (APCs) for T cells [18]. CD4⁺CD25⁺ T cells were isolated by flow cytometry and cultured in AIM V® Medium for 96 hrs with irradiated autologous APCs. RT-PCR, Western blotting and Tri-Treg cell counting were then performed.

Analysis of candidate TF binding elements disrupted by rs3761548 site alteration

The analysing tools include MAPPER [21], transcription element search system (TESS) [22], Tfsitescan [23], PROMO [24] and Genomatix-MatInspector program [25].

Cell transfection and luciferase reporter gene assay

The genomic DNA extracted from PBMCs of three psoriatic patients (HLA DR B1*07 positive) with genotype CC was used as templates. The human Foxp3 gene fragments were amplified by PCR from genomic DNA position 49008107 to 49008707 (containing core promoter [15]) and 49004666 to 49005716 (partial intron1 sequence around rs3761548). The Foxp3 amplicons (1100 bp and 600 bp) were joined and cloned into the pGL3 basic vector to generate the pGL3 FP3. Site-directed mutagenesis (C to A) in Foxp3 intron1 site 49005185 was introduced using the QuickChange (La Jolla) kit to generate pGL3 FP3M, according to the manufacturer’s instructions. All the recombinant plasmids were confirmed by DNA sequencing. CD4⁺CD25⁺ T cells were isolated and AIM V® Medium overnight and washed twice with PBS before transfection. Transfection procedures were as described with some modifications [15, 26]. Briefly, an amount of 10 μg of the pGL3 FP3 and pGL3 FP3M vectors and 1 μg of phRL-TK (as an internal control reporter) were added to 5 × 10⁶ CD4⁺CD25⁺ T cells and electroporated with appropriate parameters. After a 96-hr culture in serum-free medium with irradiated autologous APCs or phorbol myristate acetate (PMA, 20 ng/ml) and ionomycin (1 mM), luciferase activity was measured by the dual luciferase assay system, according to the manufacturer’s instructions.

Electrophoretic mobility shift assay

Nuclear extracts of CD4⁺CD25⁺ T cells were prepared as described in published studies [27, 28]. Nuclear extracts were incubated with or without anti-HEN1, anti-E47, anti-c-Myb and anti-PPAR-γ antibodies for 10 min. at room temperature, respectively. The binding reactions using these nuclear extracts were then carried out at room temperature for 20 min. in a buffer containing 20 mM Tris, pH 7.5, 50 mM NaCl, 5% glycerol, 1 mM dithiothreitol, 1 mM EDTA, 0.1–0.5 μg poly dI-dC (as the non-specific competitor), and the ³²P end-labelled probes (10⁴ cpm). For competition assays, 100-fold molar excess of unlabelled oligonucleotides was added. The binding complexes were resolved on a 5% polyacrylamide gel in 0.5 × TBE. The gels were dried and exposed to X-ray films overnight at –80°C. Films were scanned and analysed. The sequence of the double-stranded oligonucleotide probes was as follows: rs37615848 C, 5′ GCTCTCTCCCCAACTGCAGGCCTCAGTTTACCCCTCAG 3′ (top strand) and rs37615848 A, 5′ GCTCTCTCCCCAACTGAAGGCCTCAGTTTACCCCTCAG 3′ (top strand).

Chromatin immunoprecipitation assay

HLA DR B1*07 positive CD4⁺CD25⁺ T cells were stimulated for 96 hrs with irradiated autologous APCs pulsed by APL 119R (100 μg/ml). The ChIP procedures were performed according to manufacturer’s protocol with a ChIP assay kit. PCR was performed by standard protocols with Foxp3-specific primers that flanked rs3761548 site from 49005118 to 49005255 (forward, 5′-CAGAGTTGAAATCCAAGC-3′; reverse, 5′-AGAAGGGGGAATGGTAGCCCAGGTTC-3′). DNA from RNA polymerase II precipitated samples were used for glyceraldehyde-3-phosphate dehydrogenase promoter amplification. Amplicons were validated by agarose gel and DNA sequencing.

Loss of function and overexpression of E47 and c-Myb in human CD4⁺CD25⁺ T cells

Transfection of siRNAs for E47 and c-Myb to psoriatic CD4⁺CD25⁺ T cells (HLA DR B1*07 positive, allele C at rs3761548) was done at a concentration of 150 nM with HiPerFect (Qiagen) transfection reagent according to the protocol recommended. The transfections of the expression vectors (pcDNA_3-, pcDNA_3-E47, pcDNA_3-c-Myb) were performed as described in Luciferase reporter gene assay. Forty-eight hours after transfection, cells were stimulated with irradiated autologous APCs. At 24-, 48- and 96-hr points after stimulation, E47 and c-Myb expression levels and Foxp3 transcription levels were measured by Western blotting and RT-PCR assays, respectively.

Cell cycle blocker assay

A total of 1 × 10⁶ CD4⁺CD25⁺ T cells were cultured for 96 hrs in AIM V® Medium with irradiated autologous PBMCs pulsed by APL 119R. After the first 48 hrs, DMSO, Aphidicolin (1 μg/ml) or Nocodazole (100 ng/ml) were added, respectively, and cells were cultured for another 48 hrs for the induction of G1/S boundary or G2/M block. RT-PCR, Western blotting and flow cytometry analysis were performed as described above.

Statistics

The differences between groups were assessed by chi-square test for SNP analysis. Unconditional univariate and multivariate logistic regression analyses were performed to obtain the crude and adjusted odds ratios (ORs) for risk of psoriasis and their 95% confidence intervals (CIs). The differences between means were assessed by the Student’s t-test. Two-sided tests of statistical significance made at the 5% significance level were performed with the SAS software.

Results

SNP Analysis of Foxp3 gene and the correlation with clinical features of PV

Population studies show that the structural, quantitative or regulatory polymorphism at the Foxp3 locus may be contributing to the susceptibility of some autoimmune conditions [29–31]. Until now, there have been no reports about the relationship between Foxp3 gene SNPs and quantitative reduction of Tri-Treg cells in PV. We hypothesized that SNP(s) of Foxp3 gene were responsible for the reduced circulating numbers of Treg cells, imbalanced regulatory versus effector T cells ratios and clinical courses in PV. To address this, we genotyped DNA from 524 psoriatic patients and 549 matched healthy controls for SNPs of 5 candidate sites (rs3761547, rs3761548, rs3761549, rs4824747 and rs5906761).

As shown in Table 1, the Foxp3 A allele frequency at rs3761548 was significantly higher in the patients than that in healthy controls (22.2% versus 17.7%, ײ= 6.618, P= 0.010). Consistent with the A allele distribution, the frequency of the combined rs3761548 A variant genotype (CA +AA) were significantly higher in the cases (37.8%) than that in the controls (31.2%) (ײ= 4.947, P= 0.026). We found the AA genotype was associated with a higher risk to psoriasis (adjusted OR, 1.68; 95% CI, 0.97–2.92, CC genotype as the reference). We also found the combined rs3761548 A variant genotype (CA+AA) was also associated with a relatively higher risk of psoriasis (adjusted OR, 1.38; 95% CI, 1.07–1.78). There were no statistically significant difference between the psoriatic patients and controls for rs3761549, rs3761547, rs4824747 and rs5906761 sites (data not shown).

Table 1.

The results of SNP analysis for rs3761548.

Genotypes	Cases (n= 524) n	%	Controls (n= 549)*N	%	P †	Crude OR (95% CI)	Adjusted OR (95% CI)‡
rs3761548
CC	326	62.2	378	68.8		1	1
AC	163	31.1	148	27.0	0.04	1.28 (0.98–1.67)	1.32 (1.01–1.74)
AA	35	6.7	23	4.2		1.76 (1.02–3.05)	1.68 (0.97–2.92)
AC+AA	198	37.8	171	31.2	0.026	1.34 (1.04–1.73)	1.38 (1.07–1.78)
A allele	22.2		17.7	0.010

^*The observed genotype frequencies among the control patients were in agreement with the Hardy–Weinberg equilibrium (P= 0.113).

^†Two-sided chi-square test for genotype distributions and allele frequencies between the cases and controls.

^‡Odds ratios (ORs) were obtained from a logistic regression model with adjustment for age and sex; 95% CI, 95% confidence interval.

The frequency of the rs3761548 CA+AA genotypes was significantly higher in more severe psoriatic patients (PASI > 20) than that in moderate ones with PASI ≤ 20 (45.9%versus 41.0%, ײ= 9.006, P= 0.003). In the present study, we also found there was certain tendency that some autoimmune disorders (vitiligo, bullous pemphigoid, autoimmune thyroid disease, type 1 diabetes and et al.) were concomitant with psoriasis. There were significant differences among patients with variant rs3761548 genotypes. It seemed that patients with allele A at Foxp3 rs3761548 site had higher concomitance frequency and more susceptibility to those autoimmune disorders compared with those with allele C (AA versus CC, 13.7%versus 4.0%, ײ= 10.635, P= 0.001; CA+AA versus CC, 10.6%versus 4.0%, ײ= 6.839, P= 0.009). Considering the autoimmune background of psoriasis and repressive characteristics of Tri-Treg cells, we supposed that the severer clinical presentation and higher autoimmune disorder concomitance may result from variant transcription and/or expression of Foxp3 gene.

Patients with rs3761548 A had lower Foxp3 levels and impaired acceleration of Foxp3 transcription/expression after APL stimulation

As shown in (Fig. 1), resting CD4⁺CD25⁺ T cells from patients with genotype AA had lower Foxp3 transcription/expression levels than those from patients with CC in vitro (t= 4.924, P= 0.0004; t= 3.604, P= 0.004, for transcription and translation levels comparison). On the cell level, the ration of Tri-Treg cells versus T cells was also lower in patients with genotype AA (t= 3.117, P= 0.009). After APL 119R stimulation, the levels of Foxp3 transcription/expression and Treg cell levels were heightened to 12, 8 and 7 times of resting levels for patients with genotype CC, while these were highlighted to six, three and three times, respectively, for patients with AA. Based on these observations, we proposed that psoriatic patients with SNP rs3761548 A had reduced Foxp3 transcription/expression levels and impaired acceleration of transcription after APL stimulation. These findings strengthen the link between rs3761548 site and Foxp3 transcription variation.

Fig 1.

Levels of Foxp3 transcription/expression in resting and stimulated cells with different genotypes (n= 11 and 3, for CC and AA). Data were shown by mean ± S.D. and determined by the ratio between Foxp3 and β-actin for transcription (A), expression level (B) and ratio between Tri-Treg cells and lymphocytes for cell level (C) comparison. Patients with rs3761548 A had lower Foxp3 levels and impaired acceleration of Foxp3 transcription after APL stimulation. The samples were detected at equal intervals after phlebotomization (less than 3 hrs). All experiments were performed by three independent experimentalists and repeated three times. Inset, CD4⁺CD25⁺ T cells.

Allele A at SNP rs3761548 site reduced luciferase reporter activity in vitro

In order to exclude the intra-individual deviation to the fullest extent, we verified the influence of SNP rs3761548 site on the transcription levels of Foxp3 by luciferase reporter gene assay. Luciferase activity was increased six times for pGL-FP3 (allele C) and three times for pGL-FP3M (allele A) after stimulation of APL 119R and 12 times for PMA + ionomycin (not shown in the figure). The results demonstrate that the mutation from C to A at the 49005185 site dramatically reduced by 60% reporter activity (Fig. 2). Considering rs3761548 site is in close physical proximity with core promoter and elements identified pivotal for the transcription regulation [15], we proposed that the SNP rs3761548 from C to A weakened downstream Foxp3 gene transcription/expression, possibly by disrupting some TFs from binding to the element(s) adjacent to SNP rs3761548.

Fig 2.

Allele A at SNP rs3761548 site reduced reporter activity. Results given were the mean ± S.D. of luciferase light units normalized for Renilla luciferase of the same sample (n= 3). Mutation from C to A at 49005185 site dramatically reduced by 60% reporter activity All experiments were performed by three independent experimentalists and repeated three times.

Analysis of candidate TF binding elements disrupted by rs3761548 site alteration

Given the important influence of SNP rs3761548 on Foxp3 transcription in PV, we suppose that bindings of certain vital TFs to the element(s) at/adjacent rs3761548 were disrupted by this SNP. Herein, we performed a more detailed TF binding site analysis based on the allele alteration. SNP rs3761548 site locates at intron-1 of Foxp3. Published studies showed that a conserved tandem STAT binding site and a TCR-responsive enhancer overlapping a CpG island have lately been identified in Foxp3 intron-1. Compact correlations between these binding site alterations and Foxp3 expression variations were observed. All these prompt us to analyse the alteration of TF binding sites when allele C changed to A in rs3761548 in PV.

We used the mature algorithms and online resources based on pattern matching and pattern detection to identify TF binding sites at/adjacent rs3761548 in intron-1 (GeneID: 50943 in GenBank^TM). From the results of these algorithmic approaches (data not shown), we included that, in theory, at least three TF binding sites were abolished when allele C was changed to A: HEN1, E2A and c-Myb binding sites and one site was built: PPAR-γ site (Table 2).

Table 2.

Candidate TF binding elements disrupted by rs3761548 site variation analysed with MAPPER, TESS, MIRAGE, TFSEARCH, PROMO and Genomatix-MatInspector engines.

Database	Transcription factors		Brief comment
	Allele: C	Allele: A
MAPPER	HEN1*	—	HEN1: an E-box binding factor, having important roles in the development of nervous system.
		RREB-1	RREB1: detected positively in heart and kidney, enhances calcitonin and represses angiotensinogen expression.
		PPAR-γ	PPAR-γ: modulating adipocyte differentiation, glucose homeostasis and the activity of T cells.
TESS	Ker-1	—	Ker-1: controlling keratinocyte-specific gene expression.
	E2A*	—	E2A (E47 and E12): an E-box binding factor, controlling tissue-specific gene regulation of lymphoid, function as tumour suppressors in human T lineage cells.
MIRAGE	E2A*	—	E2A binding site (s01950, factor source: human beings) in allele C was abolished when in allele A.
TFSEARCH	c-Myb†	—	c-Myb: a member of Myb protooncogene family, playing specific roles in throughout haematopoietic cell and T cell functions proper.
			A hint can be gotten that there was most likely a human c-Myb binding site change when allele C changed to A, although the c-Myb (T00138, M00004) element was from specie of Mus musculus in this algorithm.
PROMO	AP-2-α A c-Myb†	AP-2-α A	The dissimilarity of c-Myb binding site (8.4439%) was more near the dissimilarity margin (15%). So we suspected that it was much more likely the c-Myb binding site if there was any TF variation around rs3761548 when allele C changed to A.
Genomatix-MatInspector	PLAG c-Myb†	PLAG	There were putative binding sites of c-Myb and PLAG in allele C around rs3761548 (RE: 1.26 and 0.90 matches per 1000 bp). In case of allele A, the c-Myb site was abolished.

^*Containing basic helix-loop-helix motifs.

†Containing helix-turn-helix motifs.

E47 and c-Myb can bind to Foxp3 rs3761548 Site in vitro and in vivo

T-cell development is guided by a complex set of TFs that act recursively, in different combinations, at each of the developmental choice points from T-lineage specification to peripheral T-cell specialization, including GATA-3, Notch/CSL, TCF-1, Ikaros, Runx, Ets, E2A/HEB and c-Myb factors [32]. The latter two factors, E2A/HEB and c-Myb, were just two of the candidate TFs regulating Foxp3 transcription based on our theoretical analysis results. Then can these factors bind to rs3761548 site in vitro and in vivo? What kind of roles these factors play in the regulation of Foxp3 transcription? We analysed these presumptions and questions using experimental methods in the following parts based on these theoretical results.

Firstly, we determined the TFs that can bind to Foxp3 rs3761548 site in intro by EMSAs. From Fig. 3A, we can see slowly migrating complexes on all the lanes except lane 1 (probe without nuclear extract) and lanes 3, 5, 7, 9 and 11 (competition by 100-fold molar excess of unlabelled oligonucleotides). Supershift assays using specific antibodies identified the E47 and c-Myb as components of the more slowly migrating complexes (arrows) on anti-E47 and anti-c-Myb lanes (4 and 6), while anti-HEN1 and anti-PPAR antibodies and an isotype-matched control antibody did not affect the mobility of the Foxp3 binding complexes. EMSA analysis of nuclear extracts prepared from CD4⁺CD25⁺ T cells by probe rs3761548A did not show any antibodies affecting the mobility of Foxp3 binding complexes (data not shown). Together, these results demonstrate that E47 and c-Myb can bind the Foxp3 sequence in vitro.

Fig 3.

E47 and c-Myb could bind to Foxp3 rs3761548 site in vitro and in invo confirmed by EMSA (A) and ChIP (B) assays (n= 3). (A) EMSA analysis of nuclear extracts prepared from CD4⁺CD25⁺ T cells by probe rs37615848C. The specific complexes were indicated with arrows. Lane 1 represents probe without nuclear extract. Lanes 3, 5, 7, 9 and 11 represent competition by 100-fold molar excess of unlabelled oligonucleotides. (B) Purified nucleoprotein complexes were obtained from CD4⁺CD25⁺ T cells via ChIP assays with anti-HEN1 (HEN1), anti-E47 (E47), anti-c-Myb (c-Myb.), anti-PPAR-γ (PPAR), mouse IgG (mIgG), goat IgG (gIgG), no antibody (Ab^–) and no chromatin (DNA^–). The precipitated DNA fractions were analysed by PCR with primers specific to the Foxp3 intron1 region around rs3761548 site. Anti-RNA polymerase II precipitated chromatin was used to analyse glyceraldehyde-3-phosphate dehydrogenase promoter as a control (not shown in the figure). The Foxp3 fragment was also amplified from the total sheared DNA (input DNA). Each EMSA and ChIP experiment had been repeated at least three times to confirm the constant and reproducible patterns shown here.

To further confirm our hypothesis that E47 and c-Myb bind the rs3761548 site specifically in invo, we carried out ChIP experiment with primers designed to amplify a 137-bp region of Foxp3 intron-1. The ChIP results (Fig. 3B) suggested that it was E47 and c-Myb, not HEN1 or PPARP, that bind to Foxp3 intron-1 in vivo (t′= 6.0529, P < 0.05; t′= 10.228, P < 0.05, compared to the value of mouse IgG, respectively). Taken together, our results further confirmed that E47 and c-Myb can bind to Foxp3 intron-1. The binding site is most likely at rs3761548 considering the different amplicon levels between genotype CC and AA and the theoretical analysis results of the E47 and c-Myb binding sites overlapping at rs37615408 site. After stimulation, both E47 and c-Myb were recruited to the Foxp3 intron-1 region containing rs3761548 site in CD4⁺CD25⁺ T cells from psoriatic patients with genotype CC, compared with those from patients with genotype AA, in which lower recruitment of E47 or c-Myb was observed (t= 4.6853, P= 0.0094; t= 5.6767, P= 0.0048, for E47 and c-Myb, respectively). These results indicated that the rs3761548 site alteration can influence the binding of E47 and c-Myb to Foxp3 intron-1 region.

E47 and c-Myb are likely to play opposite roles in the transcription regulation of Foxp3 gene

There was overlapping between E47 (CANNTG) and c-Myb (T/CAACG/TG) binding sites. It was obvious that these two factors were not likely to bind the same core sequence in rs3761548 at the same time. Then when and what kind of possible roles these two factors play?

Firstly we analysed the possible role of E47 in Foxp3 transcription. We know that TGF-β1 supports the maintenance of Foxp3 expression, regulatory function, and homeostasis in peripheral CD4⁺CD25⁺ Treg cells in mice. In human beings, high stable levels of Foxp3 expression could be induced by TCR stimulation in the presence of TGF-β, although whether this phenotype is sufficient to define a human CD4⁺ T cell as a Treg cell remains not fully defined [14, 33–35]. The mechanism of up-regulation of Foxp3 behind TCR and TGF-β stimulation is not very clear. It seems that the regulation of E protein activity by TCR stimulation and TGF-β signalling plays pivotal roles in this regulation. According to published literatures [36, 37], TCR-mediated signalling or TGF-β1 treatment resulted in up-regulation of Id3 protein (an inhibitor helix-loop-helix (HLH) protein) and a concomitant down-regulation of DNA binding activity of E47-HEB TFs (Fig. 4). We suppose that E47-HEB maybe plays repressive role in the transcription of Foxp3 gene. According to our above SNP results that the transcription level of Foxp3 gene was lower in mutated individuals than that with wild genotype (CC), similar difference was observed after stimulation by APL 119R, so it seemed that c-Myb most likely played an active role in regulation of Foxp3 transcription.

Fig 4.

TCR-Id3 and TGF-β-Id3 pathway decreasing the affinity of E47/HEB DNA binding in lymphocytes. TCR-mediated signalling modulates E47-HEB DNA-binding activity by activation of Id3 expression, which is mediated by Erk MAP kinase activation and the transcriptional regulators SAP-1 and Egr-1. Similarly, signalling mediated by TGF-β activates Id3 expression to modulate E47 and/or HEB DNA binding by direct activation of SMAD. Id3 down-regulates DNA binding activity of E47-HEB transcription factors. LAT, linker of activation in T cells; Zap70, ζ-chain-associated protein; PLC, phospholipase C; SAP-1, stress-activated protein and Egr1, early growth response gene 1. Adapted from [36–38].

We further confirmed above theoretical analysis by performing siRNA and overexpression assays. We found that E47 expression and Foxp3 transcription levels increased and decreased oppositely in CD4⁺CD25⁺ T cells. E47 expression level was raised (compared with pcDNA_3- vector) and Foxp3 transcription level was decreased by transfection of pcDNA_3-E47 (t= 4.930, P= 0.008; t= 4.509, P= 0.011, compared with pcDNA_3- vector, respectively). After transfection of siRNA for E47, E47 expression level was blocked and Foxp3 transcription level was increased (t= 1.990, P= 0.118; t= 2.488, P= 0.068, compared with control siRNA, respectively), although not statistically significant (Fig. 5A). For c-Myb, its expression and Foxp3 transcription level changed concordantly. c-Myb expression level was raised and Foxp3 transcription level was increased by transfection of pcDNA_3- c-Myb (t= 3.156, P= 0.034; t= 3.379, P= 0.028, compared with pcDNA_3- vector, respectively). By transfection of siRNAs for c-Myb, its expression level was blocked and Foxp3 transcription level was decreased (t= 3.670, P= 0.021; t= 3.087, P= 0.019, compared with control siRNA, respectively) (Fig. 5B).

Fig 5.

Evaluation of the impacts of E47 (A) and c-Myb (B) on the Foxp3 transcription by siRNA and overexpression assays. #1 E47 expression level was raised by transfection of pcDNA_3-E47, compared with blank medium and pcDNA_3- vector (t= 4.312, P= 0.012; t= 4.930, P= 0.008, respectively). #2 Foxp3 transcription level was decreased after transfection of pcDNA_3-E47, compared with blank medium and pcDNA_3- vector (t= 4.621, P= 0.010; t= 4.509, P= 0.011, respectively). #3 E47 expression level was blocked by transfection of siRNA for E47, compared with blank medium and control siRNA (t= 1.755, P= 0.154; t= 1.990, P= 0.118, respectively, not statistically significant). #4 Foxp3 transcription level was increased after transfection of siRNA for E47, compared with blank medium and control siRNA (t= 2.038, P= 0.111; t= 2.488, P= 0.068, respectively, not statistically significant). #5 c-Myb expression level was raised by transfection of pcDNA_3-c-Myb, compared with blank medium and pcDNA_3- vector (t= 3.002, P= 0.040; t= 3.156, P= 0.034, respectively). #6 Foxp3 transcription level was increased after transfection of pcDNA_3-c-Myb, compared with blank medium and pcDNA_3- vector (t= 3.948, P= 0.017; t= 3.379, P= 0.028, respectively). #7 c-Myb expression level was blocked by transfection of siRNAs for c-Myb, compared with blank medium and pcDNA_3- vector (t= 2.871, P= 0.045; t= 3.670, P= 0.021, respectively). #8 Foxp3 transcription level was decreased after transfection of siRNAs for c-Myb, compared with blank medium and pcDNA_3- vector (t= 3.151, P= 0.035; t= 3.087, P= 0.019, respectively). All experiments were performed by three independent experimentalists and repeated three times.

As for the insignificantly decreased E47 expression and increased Foxp3 transcription levels by E47 siRNA, we think it may due to the early expression window of E47 (mainly at G1 phase according to cell cycle analysis in (Fig. 6)) and its short half-life [39]. So we also measured E47 expression and Foxp3 transcription levels in the loss of function and overexpression assays at 24 hrs after stimulation of autologous APCs pulsed by APL 119R (Fig. 7). We found that E47 expression and Foxp3 transcription levels increased and decreased oppositely and significantly in CD4⁺CD25⁺ T cells at early cycle phase.

Fig 6.

Representative figure demonstrating the alterations of E47, c-Myb, cyclins and Foxp3 at different cycle phases of CD4⁺CD25⁺ T cells with (D, E) or without (A, B, C) Nocodazole or Aphidicolin block. HLA DRB1*07 + CD4⁺CD25⁺ T cells from psoriatic patients with genotype CC (n= 4) and AA (n= 3) were cultured in AIM V® Medium with irradiated autologous APCs for 48 hrs, followed by another 48-hr culture with or without Nocodazole or Aphidicolin. Cells were collected for Western blotting (B, E) and RT-PCR (A, D) analysis at 0-, 24-, 48- and 96-hr points, respectively. Dynamic fitted curves (C) were made according to A and B. The target proteins were normalized to β-actin and Foxp3 gene transcription level was also normalized to β-actin gene. All experiments were performed by three independent experimentalists and repeated three times.

Fig 7.

Evaluation of the early impact of E47 on the Foxp3 transcription by siRNA and overexpression assays. siRNAs and expression vectors for E47 and c-Myb were transfected into psoriatic CD4⁺CD25⁺ T cells (n= 3, HLA DR B1*07 positive, allele C at rs3761548). #1 E47 expression level was raised by transfection of pcDNA_3-E47, compared with blank medium and pcDNA_3- vector (t= 3.177, P= 0.034; t= 3.051, P= 0.038, respectively). #2 Foxp3 transcription level was decreased after transfection of pcDNA₃-E47, compared with blank medium and pcDNA3- vector (t= 3.361, P= 0.028; t= 8.251, P= 0.001, respectively). #3 E47 expression level was blocked by transfection of siRNA for E47, compared with blank medium and control siRNA (t= 3.292, P= 0.030; t= 3.486, P= 0.025, respectively). #4 Foxp3 transcription level was increased after transfection of siRNA for E47, compared with blank medium and control siRNA (t= 3.291, P= 0.030; t= 5.044, P= 0.007, respectively).

E47 down-regulates Foxp3 transcription mainly at G1 and c-Myb plays opposite role at G2/M phase

Then as we analysed above, these two factors were not likely to bind the same core sequence at rs3761548 at the same time. When do these two factors play their roles, respectively? According to published literature, E47 co-ordinately regulates the expression of genes involved in cell cycle progression, also as targets of G1 cyclin-dependent Kinases, mainly in G1 [40]. c-Myb contributes to G2/M and G1/S cell cycle transition in human lymphocytes [41, 42]. Are these true for human peripheral CD4⁺CD25⁺Treg? We designed following experiments to detect cell cycles and analyse the levels of E47, c-Myb, cyclins and Foxp3 at different phases of cell cycle (with/without cell cycle blockers) by RT-PCR and/or Western blotting.

We found that c-Myb expression and cyclin B1 expression changed concordantly in HLA DRB1*07 + CD4⁺CD25⁺ T cells. E47 expression, cyclin D1 and cyclin E expression also showed the similar relationship. For the cells from patients of genotype CC, increased level of E47 and decreased Foxp3 transcription levels at 24 hrs (approximately G1 phase) were observed. At 96 hrs (approximately G2/M phase), increased c-Myb and high levels of Foxp3 transcription can be found. On the contrary, as for cells from genotype AA, increased levels of E47/c-Myb were concordant with high/lower Foxp3 transcription levels at 24-/96-hr points (Fig. 6A, B and C). These variations were highly because of the disruption of E47/c-Myb binding site by rs3761548 C to A. These results further confirmed the activator/repressor roles of E47/c-Myb on the transcription of Foxp3 gene.

Nocodazole, which interrupts the microtubule assembly and blocks the cells at M phase (with increased expression of c-Myb and cyclin B1, reduced cyclin D1 and cyclin E), induced higher levels of Foxp3 transcription and expression in CD4⁺CD25⁺ T cells from psoriatic patients with genotype CC. Similar results were seen with Taxol, another G2/M blocking agent (data not shown). But as for the cells of genotype AA, although accumulation of c-Myb after Nocodazole block at G2/M, increased levels of Foxp3 transcription was not observed, which maybe result from the disruption of c-Myb binding site by the SNP. CD4⁺CD25⁺ T cells were arrested at the G1/S boundary by being subjected to Aphidicolin, an inhibitor of eukaryotic nuclear DNA replication. It led to increased expression of E47, cyclin D1, cyclin E, and reduced cyclin B1 and Foxp3 transcription/expression. Similar to the opposite phenomena observed in the Nocodazole block for c-Myb between cells from genotype CC and AA, high levels of E47 by Aphidicolin failed to induce lower transcription of Foxp3 in cells of genotype AA compared with that in genotype CC (Fig. 6D and E). The mechanism behind was also most likely involved with the disruption of E47 binding site by the SNP rs3761548.

Based on these results, we proposed that E47/c-Myb responsive regulatory elements were present in the intron-1 of Foxp3 gene. E47 and c-Myb bound to these overlapping elements (rs3761548) to oppositely regulate the transcription of Foxp3 mainly at G1 and G2/M phase in cell cycle, respectively. For some psoriatic patients with the SNP rs3761548 A, binding of E47 and c-Myb to this element is disrupted by the SNP, leading to decreased transcription levels of Foxp3 gene.

Discussion

In this study, we performed large-scale SNP investigation in psoriatic patients and found a potent regulation site, rs3761548, in intron-1 that was correlated with Foxp3 transcription level variations in psoriatic patients with different genotypes. Patients with rs3761548 AA have severer clinical presentation, higher autoimmune disorder concomitance, reduced Foxp3 levels and defective acceleration of Foxp3 transcription/expression after APL stimulation, compared with those of genotype CC. These observations indicate the relationship among rs3761548 SNP, variant Foxp3 transcription levels and autoimmune disease susceptibility.

The strengthened link between rs3761548 site and Foxp3 transcription variation suggested that the SNP rs3761548 from C to A weakened downstream Foxp3 gene transcription, most likely by disrupting some TFs from binding to this site. Based on these assumptions, we analysed and demonstrated potent new E47-/c-Myb-dependent regulation elements overlapping at rs3761548 site (CANNTG, T/CAACG/TG for E47 and c-Myb, respectively) in patients with genotype CC. E47 repressed Foxp3 gene transcription at cell cycle of G1 phase and c-Myb promoted the transcription at G2/M phase of Treg in PV. For some psoriatic patients with the SNP rs3761548 A, bindings of E47 and c-Myb to their elements were disrupted by the SNP, leading to decreased transcription levels of Foxp3 gene. Our findings also demonstrate that not only may SNPs within coding sequences and proximal promoter regions of genes be functionally important, but SNP(s) in introns might also be critical for transcriptional regulation of Foxp3, and thus the development of PV.

E47 is a member of class I basic HLH (class I bHLH, also known as E proteins) TF family. E proteins contain an HLH dimerization domain and a basic region that mediates DNA binding with sequences that contain CANNTG, referred to as an E-box [43]. E47 co-ordinately regulates the expression of genes involved in cell survival and lymphoid maturation, function as tumour suppressors in human T lineage cells [44, 45]. The expression pattern of E47 during lymphocyte development is very dynamic. E47 expression is highest in the DN (double negative) compartment, decreases during the transition toward the DP (double positive) cell stage, can barely be detected in single-positive cells and increases again in activated T lineage cells. In the present study, we demonstrated that E47 played repressive role in the transcription of Foxp3 gene, mainly at G1 phase of CD4⁺CD25⁺ T cells. But it does not necessarily mean that the pathway of TCR-Id3 and TGF-β-Id3 also exist in Tri-Treg cells, like in thymocytes and lymphocytes, which need to be further investigated.

When at G2/M phase of CD4⁺CD25⁺ T cells, E47 decreased and c-Myb was up-regulated with elevated levels of Foxp3 gene in CD4⁺CD25⁺ T cells. c-Myb is a member of Myb protooncogene family and contains two consecutive helix-turn-helix motifs with unconventional turns. These two helix-turn-helix motifs which are similar with the HLH motif of E proteins to some extent and are absolutely required for complex formation of Myb with DNA. c-Myb specifically binds to T/CAACG/TG sequences with some distinct preferences for the nucleotides flanking this core binding site [46]. c-Myb plays specific roles in throughout T cell functions proper, in addition to be required during the multiple stages of B-lymphopoiesis [47]. c-Myb is directly involved in the formation of CD4⁺CD8⁺CD25⁺, CD4⁺CD25⁺ and CD8⁺CD25⁺, in which Foxp3 mRNA were positive. c-Myb is also important for the proliferative responses of mature CD4 and CD8 T cells in the periphery [48]. All these processes may imply an important role for c-Myb in autoimmune dysfunction.

Besides their functions in normal T-cell development and differentiation, evidence suggested that E47 and c-Myb are also involved in the development of T-cell lymphoma or T-cell leukaemia [45, 49, 50], in which Foxp3 mRNA transcriptions were also found positively. Evidence indicates that Tri-Treg maintain peripheral tolerance to self-antigens and also inhibit anti-tumour immune responses [51]. The expression of Foxp3 in such tumour cells suggests its important roles in the suppression of host anti-tumour immune response. Manipulating the transcription of Foxp3 maybe one of the mechanisms of E47 and c-Myb in tumour differentiation. All these clues facilitate our conclusion that E47 and c-Myb participate in the regulation of Foxp3 transcription.

Taken together, these studies demonstrate that E47 and c-Myb are necessary for transcriptional regulation of Foxp3. Are they the sole initial factors for the activation of Foxp3 transcription? It seems not the truth. We know that T-cell development involves successive waves of regulated gene expression. The composition of functional TFs can shift with each developmental transition, whereas the regulatory elements for each gene are fixed through developmental time and space [52]. In some type cells, especially in human lymphatic system, some Foxp3-negetive cells also contain high-level E47/c-Myb factors and wild-type rs3761548 site of Foxp3 DNA. So we propose that E47 and c-Myb factors are not the sole initial factors for the activation of Foxp3 transcription and they most likely work cooperatively with other factors binding in close physical proximity, which remains to be investigated.

According to data from published studies, there appears to be more than one pathway to the regulation of Foxp3 transcription (Fig. 8). Mantel et al. identified the basal promoter of Foxp3 gene containing six NF-AT and AP-1 binding sites, which were positively regulating the trans activation of the Foxp3 promoter after triggering of the TCR [15]. The effects of cytokine IL-2 and IL-4 signalling pathways in human Treg have also been examined. IL-2 selectively up-regulated the expression of Foxp3 in purified CD4⁺CD25⁺ T cells but not in CD4⁺CD25^– cells in vitro. This regulation involved the binding of STAT3 and STAT5 proteins to highly conserved STAT binding sites located in the first intron of the Foxp3 gene [16]. IL-4 inhibited the TGF-β-mediated Foxp3 induction. This inhibition is mediated by direct binding of IL-4-induced GATA3 to the Foxp3 promoter [53]. Quintana et al. reported the identification of the ligand-activated transcription factor aryl hydrocarbon receptor as a regulator of Foxp3 and Treg cell differentiation in mice. Aryl hydrocarbon receptor activation by its ligand 2,3,7,8-tetrachlorodibenzo-p-dioxin induced functional Treg cells that suppressed experimental autoimmune encephalomyelitis [54]. DNA demethylation is another hot spot in the study of gene transcription. Only demethylation at the Foxp3 locus was found to be restricted to Treg when tested against all major peripheral blood cell types and a selection of non-blood cells. Most importantly, Foxp3 demethylation was only observed in natural Treg (including upon in vitro expansion), but not in activated conventional T cells [55]. Recently, a TCR-responsive enhancer in the Foxp3 intron-1 has been identified. The enhancer is dependent on a cyclic-AMP response element binding protein (CREB)/activating transcription factor (ATF) site overlapping a CpG island. Methylation of this island inversely correlates with CREB binding and Foxp3 expression. TGF-β1, which induces Treg cell formation, decreases methylation of the CpG island and increases Foxp3 expression [17]. Floess et al. identified an evolutionarily conserved region within the Foxp3 locus in intron-1 (about from 49004600 to 49004000, containing CREB/ATF binding site) possessing transcriptional activity. Bisulphite sequencing and chromatin immunoprecipitation revealed complete demethylation of CpG motifs as well as histone modifications within the conserved region in ex vivo isolated Tri-Treg cells in mice. Treg induced by TGF-βin vitro display incomplete demethylation of this locus [56].

Fig 8.

Schematic view on the regulation pathways of Foxp3 transcription based on published data and our results. (1) IL-4–GATA3 pathway. Species: human beings and mice. Data from [53]. (2) TCR- basal core promoter pathway. Species: human beings. Data from [15]. (3) and (6) TCR-Id3-E47, TGF-β-Id3-E47, TCR- c-Myb, and TGF-β- c-Myb pathways, identified in thymocytes and lymphocytes, but needed to be investigated in Treg cells. Species: human beings. The binding sites of E47/c-Myb on Foxp3 intron-1 are 490051856 nearby. Data from our results, [36, 37]. (4) TCR-CREB/ATF pathway. Species: mice. The binding site of CREB/ATF on Foxp3 intron-1 is 49004196 nearby. Data from [17]. (5) IL-2 -STAT pathway. Species: human beings. The binding sites of STAT3/ STAT5 on Foxp3 intron-1 are 49004086 and 49004286 nearby. Data from [16]. (7) TGF-β-demethylation pathway. TGF-β induced incomplete demethylation for induced Treg cells. Species: mice. The demethylation region for TGF-β is from 49004000 to 49004600 nearby. Data from [17, 54]. (8) Ligand 2,3,7,8-tetrachlorodibenzo-p-dioxin induced Foxp3 via its aryl hydrocarbon receptor. Species: mice. Data from [54].

In this study, we demonstrate that E47 and c-Myb can bind to potent new elements (490051856 nearby) at G1 and G2/M windows, respectively, and played opposite roles in the regulation of Foxp3 transcription. According to data from published studies [36, 37], we proposed that TCR-Id3-E47, TGF-β-Id3-E47, TCR- c-Myb and TGF-β- c-Myb pathways, already identified in thymocytes and lymphocytes, may also exist in Treg cells, but needed to be further investigated. More importantly, whether these E47/c-Myb-dependent regulatory elements remain active in psoriatic patients with other HLA types? Is it also responsible for defective transcription of Foxp3 in other autoimmune disorders? What are the relationships among the pathways, also the relationships between the involved TFs, described in (Fig. 8)? What do the complete sets of binding factors necessary for Foxp3 transcriptional activation consist of? Are E47 and c-Myb the members in the transcription complex and what are their orientations in the complex? All these are pivotal for the clarification of Foxp3 transcription regulation and are urgent to be interpreted.

In conclusion, we demonstrated potent new TCR response and E47-/c-Myb-dependent regulation elements in the first intron of Foxp3 gene based on large-scale SNP analysis. E47 and c-Myb antagonize each other at this overlapping binding site to oppositely regulate the transcription of Foxp3 at G1 and G2/M phase, respectively. For patients with genotype AA, the polymorphism causes loss of bindings to the E47 and c-Myb factors, leading to defective transcription of Foxp3 gene. This is the first example of loss E47/c-Myb binding site associated with susceptibility to an autoimmune disease. Further identification of the networks and molecular mechanisms underlying Foxp3 transcription may provide new insights into Foxp3 transcriptional regulation and alternative therapeutic strategies to improve characteristics of autoimmune disorders.

Conflict of Interest Disclosure

None declared.