Valproate Treatment of Human Cord Blood CD4-positive Effector T Cells Confers on Them the Molecular Profile (MicroRNA Signature and FOXP3 Expression) of Natural Regulatory CD4-positive Cells through Inhibition of Histone Deacetylase*

Regulatory T cells (Tregs) play a key role in immune system homeostasis and tolerance to antigens, thereby preventing autoimmunity, and may be partly responsible for the lack of an appropriate immune response against tumor cells. Although not sufficient, a high expression of forkhead box P3 (FOXP3) is necessary for their suppressive function. Recent reports have shown that histones deacetylase inhibitors increased FOXP3 expression in T cells. We therefore decided to investigate in non-Tregs CD4-positive cells, the mechanisms by which an aspecific opening of the chromatin could lead to an increased FOXP3 expression. We focused on binding of potentially activating transcription factors to the promoter region of FOXP3 and on modifications in the five miRs constituting the Tregs signature. Valproate treatment induced binding of Ets-1 and Ets-2 to the FOXP3 promoter and acted positively on its expression, by increasing the acetylation of histone H4 lysines. Valproate treatment also induced the acquisition of the miRs Tregs signature. To elucidate whether the changes in the miRs expression could be due to the increased FOXP3 expression, we transduced these non-Tregs with a FOXP3 lentiviral expression vector, and found no changes in miRs expression. Therefore, the modification in their miRs expression profile is not due to an increased expression of FOXP3 but directly results from histones deacetylase inhibition. Rather, the increased FOXP3 expression results from the additive effects of Ets factors binding and the change in expression level of miR-21 and miR-31. We conclude that valproate treatment of human non-Tregs confers on them a molecular profile similar to that of their regulatory counterpart.


Regulatory T cells (Tregs) play a key role in immune system homeostasis and tolerance to antigens, thereby preventing autoimmunity, and may be partly responsible for the lack of an appropriate immune response against tumor cells. Although not sufficient, a high expression of forkhead box P3 (FOXP3) is necessary for their suppressive function. Recent reports have shown that histones deacetylase inhibitors increased FOXP3 expression in T cells.
We therefore decided to investigate in non-Tregs CD4-positive cells, the mechanisms by which an aspecific opening of the chromatin could lead to an increased FOXP3 expression. We focused on binding of potentially activating transcription factors to the promoter region of FOXP3 and on modifications in the five miRs constituting the Tregs signature. Valproate treatment induced binding of Ets-1 and Ets-2 to the FOXP3 promoter and acted positively on its expression, by increasing the acetylation of histone H4 lysines. Valproate treatment also induced the acquisition of the miRs Tregs signature. To elucidate whether the changes in the miRs expression could be due to the increased FOXP3 expression, we transduced these non-Tregs with a FOXP3 lentiviral expression vector, and found no changes in miRs expression. Therefore, the modification in their miRs expression profile is not due to an increased expression of FOXP3 but directly results from histones deacetylase inhibition. Rather, the increased FOXP3 expression results from the additive effects of Ets factors binding and the change in expression level of miR-21 and miR-31. We conclude that valproate treatment of human non-Tregs confers on them a molecular profile similar to that of their regulatory counterpart.
Tregs exist in two general categories: thymus-derived Tregs, referred to as natural Tregs (nTregs), and adaptive Tregs, which can be induced in the periphery in response to a variety of stimuli, including pathogens, IL-10, and transforming growth factor-␤ (11)(12)(13)(14)(15), or in vitro from naïve CD4ϩ, CD25Ϫ T cells upon activation of the T cell receptor in combination with transforming growth factor-␤ treatment (16 -18). In addition to a distinct combination of membrane markers, the hallmark of nTregs cells is their high expression level of the transcription factor forkhead box P3 (FOXP3), which is indispensable to their suppressive activity, phenotype, stability, and survival in the periphery (3). Mutations of the FOXP3 gene lead to the lymphoproliferative disease of the Scurfy mouse and the homologous autoimmune lymphoproliferative disorder in man, termed immune deregulation polyendocrinopathy enteropathy X-linked syndrome (19). The importance of FOXP3 for Tregs cell function is supported by the observation that its ectopic expression in effector T cells endows them with regulatory properties and some but not all of the phenotypic markers of regulatory T cells (20,21). Therefore, the understanding of the mechanisms regulating the expression of FOXP3 is of utmost importance to get an insight into pathological conditions, such as certain autoimmune diseases, or the impaired immune antitumor response linked with Tregs generation.
Histone acetylation plays a key role in transcriptional regulation (22), probably by altering chromatin structure. Acetylation of core nucleosomal histones is regulated by the balance between the activity of histones acetyltransferases and histones deacetylases (HDACs). Histones acetyltransferases preferentially acetylate specific histone lysine substrates (23,24) thus neutralizing the lysine residues positive charge and disrupting the nucleosome structure, allowing unfolding of the associated DNA, access by transcription factors, and changes in gene expression. In contrast, HDACs restore the positive charge on lysine residues through the removal of the acetyl group from lysine residues of some histones, causing the repression of gene transcription by compacting the chromatin structure (25).
Pharmacological agents that inhibit HDACs impede the removal of acetyl groups thus facilitating transcription of genes so repressed (26). In particular, HDAC inhibitors induce expression of epigenetically silenced genes that could promote growth arrest (27,28), cell differentiation (29 -31), and death in tumors of various lineages (28,(32)(33)(34). In addition, these inhibitors are effective in promoting the transcription of genes that are silenced due to CpG island hypermethylation (35)(36)(37). FOXP3 is considered a potential target for the action of HDAC inhibitors, because studies of the promoter and the surrounding chromatin have revealed histone H4 hyperacetylation when the gene was activated (38). In addition, a recent study (39) showed that treatment of human CD4ϩ, CD25Ϫ T cells with the HDAC inhibitors MS-275 and SAHA induced FOXP3 expression.
Valproic acid (2-propylpentanoic acid), a member of branched short chain fatty acids known to induce differentiation in neuroblastoma cells and inhibit their growth (40), in addition to inducing apoptosis in selected solid tumors as well as in hematological neoplasia (41,42), has been shown to be a histone deacetylase inhibitor that binds to and directly inhibits HDAC (43,44) thus inducing histone acetylation (43)(44)(45), DNA demethylation (46), chromatin decondensation (47), and the aspecific expression of a variety of genes (48 -50). It also has the potential advantage of being usable in vivo, having been used for decades to treat some forms of epilepsy, at doses compatible with HDAC inhibition (51). We therefore focused on mechanisms possibly involved in the epigenetic regulation of FOXP3 expression by HDAC inhibitors, in the human CD4-positive CD25-negative T cells model that we previously described (52).
Among possible mechanisms initiated by inhibition of HDACs, we chose to focus on potential, but yet undescribed transcription factors binding to the FOXP3 promoter region, rendered accessible by histone acetylation, as well as on the possible modulation of the microRNA signature, using valproate as an HDAC inhibitor, at concentrations close to the upper limit of pyramidal syndromes therapeutic window.
In our study, we have identified two evolutionarily conserved Ets binding sites in the human FOXP3 promoter region and demonstrated the in vivo binding of Ets-1 and Ets-2 to these sites, the latter only after chromatin opening in CD4ϩ, CD25Ϫ T cells, and their positive effect on FOXP3 expression. Treatment of human CD4ϩ, CD25Ϫ T cells with sodium valproate resulted in the induction of FOXP3 in a time-dependent man-ner, associated with the binding of Ets-1 and Ets-2 to the promoter, and with the change of expression of miR-21 and miR-31, previously shown to regulate FOXP3 expression (52), and importantly, albeit unexpectedly, with the transient acquisition of the nTregs microRNA signature. Furthermore, we were able to demonstrate that the changes in the miRs expression profile was not due to an increased expression of FOXP3, but rather this augmented FOXP3 expression could be due to the additive effects of Ets factors binding to its promoter, via the opening of the chromatin, and the changes in the expression levels of miR-21 and miR-31.

MATERIALS AND METHODS
Bioinformatics-Genomic sequences spanning the 5Ј-untranslated region of the FOXP3 gene were analyzed using the alignment software ClustalW (available online) allowing the identification of conserved regions. Transcription factor binding sites were identified using TESS software (available online).
Collection and Preparation of UCB Samples-After informed consent, umbilical cord blood (UCB) was taken from the umbilical vein after normal full-term deliveries. UCB mononuclear cells were isolated by centrifugation over a lymphocyte separation medium (PAA Laboratories, Linz, Austria).
Isolation of T-cell Subpopulations-UCB CD4ϩ lymphocytes were purified using the CD4ϩ T cells isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer's protocol. Briefly, UCB mononuclear cells were first incubated in phosphate-buffered saline supplemented with 2% decomplemented fetal bovine serum and saturating amounts of a biotin-conjugated antibodies mixture (CD8, CD14, CD16, CD19, CD36, CD56, CD123, T cell receptor ␥/␦, and Glycophorin A). Cells were then incubated with anti-biotin micro-beads, and CD4ϩ T cells were negatively selected using magnetic separation columns (Miltenyi Biotec). The CD4ϩ population was then incubated with anti-CD25 micro-beads, and the CD25ϩ and CD25Ϫ cells were sorted using magnetic columns (Miltenyi Biotec).
Cell Culture and Valproate Treatment of Human CD4ϩ, CD25Ϫ UCB T Cells-CD4ϩ, CD25ϪT cells were plated at a density of 2 ϫ 10 6 cells/well in 12-well tissue culture plates in 1 ml of RPMI 1640 supplemented with 10% AB serum, 2 mM L-glutamine, 50 units/ml penicillin, 50 g/ml streptomycin (Lonza Europe, Verviers, Belgium), in the presence of 5 g/ml phytohemagglutinin (Sigma-Aldrich) and 20 units/ml interleukin-2. The plates were incubated at 37°C in a humidified atmosphere containing 5% CO 2 for 18 -24 h. The next day, the cells were treated with valproate at a concentration of 1 mM and sampled at 2, 4, 6, and 24 h for further analyses.
Quantitative PCR for FOXP3 mRNA Expression-Total RNA was extracted with the TRIzol reagent according to the manufacturer's guidelines (Invitrogen), and first strand cDNAs were synthesized by reverse transcription using the Superscript First-strand Synthesis System as described for the RT-PCR kit (Invitrogen). Quantitative mRNA expression was measured by real-time PCR, using the Prism 7900 sequence detection system (PE, Applied Biosystems, Foster City, CA), and the TaqMan Master mix kit. EF1-␣ was used as an internal control. The FOXP3 primers and the internal fluorescence TaqMan probe were designed as follows: forward, 5Ј-TTTCACCTACGCCA-CGCTCA-3Ј; reverse, 5Ј-CCAGCTCATCCACGGTCCA-3Ј; and probe, 5Ј-FAM-CCACCTGGAAGAACGCCATCCGC-TAMRA-3Ј.
Western Blot Analysis-4 ϫ 10 6 cells were lysed and subjected to SDS-PAGE using 10% polyacrylamide gels and transferred to polyvinylidene fluoride membranes (Amersham Biosciences) using a semidry electroblot chamber. Membranes were blocked with TBST containing 5% bovine serum albumin overnight at 4°C. After blocking, the blots were incubated with a 1/200 dilution of goat anti-FOXP3 antibody diluted in TBST for 1 h at 25°C. Following 1 h of incubation with anti-goat peroxidase-conjugated antibody (Sigma) at room temperature, proteins were detected by the electrogenerated chemiluminescence method (Amersham Biosciences), according to the manufacturer's instructions.
To confirm sample loading and transfer, membranes were incubated in stripping buffer and reblocked for 1 h, and then reprobed using anti-Actin (C-2, Santa Cruz Biotechnology). All protein signals were visualized using the LAS-3000 image reader (Fuji), and signals were analyzed with AIDA software (Raytest).
Detection of Mature microRNAs by TaqMan Real-time PCR-TaqMan miR assays (ABI, Forest City, CA) used the stem-loop method (53,54) to determine the expression level of mature microRNAs. For RT reactions, 10 ng of total RNA was used in each reaction (15 l) and mixed with the RT primer (3 l). The RT reaction was carried out in the following conditions: 16°C for 30 min, 42°C for 30 min, 85°C for 5 min, and then kept at 4°C. After the RT reaction, the cDNA products were diluted five times, and 9 l of the diluted cDNA was used for PCR reaction along with TaqMan Microarray assay (1 l) and PCR master mix (10 l). The PCR reaction was conducted at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 60 s in the ABI 7900 real-time PCR system. The real-time PCR results were analyzed and expressed as relative miR expression of the CT (threshold cycle) value, which was then converted to -fold changes. RT primers, PCR primers, and TaqMan probes for miR-21, miR-31, miR-125a, miR-181c, and miR-374 were purchased from ABI. RNU44 was used for normalization.
Plasmid Constructions and Transfection Experiments-A 600-bp fragment encompassing the region upstream of the human FOXP3-translated region was cloned upstream of the Firefly luciferase gene (SacI/HindIII sites) in the PGL3 plasmid (Promega, Madison, WI) and designated as PGL3-F-WT. The PCR primers used for amplification of this region were (5Ј to 3Ј ends): forward primer, AAAAAAATTTGGATTATTAG-AAGA; and reverse primer, TGTGGTGAGGGGAA-GAAATCATAT.
Transient Transfection-Jurkat cells (3 ϫ 10 6 ) were transiently transfected using standard DEAE-dextran protocols. Cells were washed once with STBS 1ϫ (25 mM Tris-HCl (pH 7.5), 1.37 mM NaCl, 5 mM KCl, 500 M CaCl 2 , 500 M MgCl 2 , and 600 M Na 2 HPO 4 ) and resuspended in 410 l of STBS 1ϫ containing 450 g/ml DEAE-dextran, 0.5 g of the reporter plasmid, and 0.06 g of an internal control plasmid containing the Renilla luciferase gene under control of the herpes simplex virus-1 thymidine kinase promoter (pRL-TK vector, Promega). In cotransfection experiments, an additional 0.085 g of the p-GL3-F-Ets was added with the reporter plasmid constructs. Luciferase was detected using a dualluciferase reporter assay system (Promega), according to the manufacturer's instructions.
ChIP Assay and Quantitative Real-time PCR-The chromatin immunoprecipitation (ChIP) assay was performed using the kit purchased from Upstate Biotechnology following the manufacturer's protocol. Cells were fixed with 1.5% formaldehyde for 10 min at 37°C. Chromatin was isolated, sheared using a Bioruptor (Diagenode), and immunoprecipitated with Abs directed to Ets-1 (catalogue no. sc-55581 X), Ets-2 (catalogue no. sc-22803 X), PU.1 (catalogue no. sc-352 X) (all from Santa Cruz Biotechnology), acetylated histone H4 (Upstate Biotechnology, catalogue no. 06-866), HDAC (Upstate Biotechnology, catalogue no. 17-245) or control rabbit IgG (Upstate Biotechnology, catalogue no. 12-370). Cross-linking was reversed by heating, and the proteins were removed subsequently by proteinase K digestion. The presence of selected DNA sequences in the immunoprecipitated DNA was assessed by PCR using the following primer pairs (5Ј to 3Ј ends): forward, TGAGCCCTATTATCTCATTG; and reverse, CCGTT-TAAGTCTCATAATCA.
The amplified 32 P-labeled PCR products were separated on a 6% acrylamide gel and detected by autoradiography. The IL-12p35 gene promoter was used as negative control with its own primers sets (not specified here).
Real-time PCR was performed in triplicate using the Prism 7900 sequence detection system (PE Applied Biosystems). Quantitative PCR reactions were performed under conditions standardized for each primer set. Each PCR reaction was carried out in duplicate in a 20-l reaction mixture: 5 l of the eluted immunoprecipitated DNA and the SYBR Green Master mix kit (Applied Biosystems).
Dissociation curves were analyzed as a means to ensure the quality of amplicon and to monitor primer dimers. Enrichment was determined based on critical threshold (C T ) measurements (changes in fluorescence per PCR cycle number at a given threshold). The amount of genomic DNA coprecipitated with specific antibody was calculated in comparison to the total input DNA used for each immunoprecipitation in the following way: ⌬C T ϭ C T (genomic input) Ϫ C T (specific antibody), where C T (genomic input) and C T (specific antibody) are the mean threshold cycles of PCR performed in duplicates on DNA samples from the genomic input samples and the specific antibody samples, respectively. 2 ⌬CT target values were calculated for each antibody.

Lentiviral Particles Production and Transduction of Non-Tregs T Cells-Precision LentiORF pLOC lentiviral vectors purchased from Open Biosystems are lentiviral-based vectors
in which the open reading frame (ORF) of the gene of interest has been cloned downstream the CMV promoter and contain tGFP as a reporter gene. The production of VSV-G pseudotyped lenti-ORF viral particles was performed as previously described in (52).
After the optimization of transduction conditions using CD4ϩ T cells, we defined an MOI of five as the one providing the highest transduction rate. This MOI was then chosen to transduce purified CD4ϩ, CD25Ϫ UCB T cells.
Twenty-four hours after the purification procedure, the cells were exposed to lentiviral vector preparations (multiplicity of infection ϭ 5) in a volume of 500 l containing 8 g/ml Polybrene (Sigma). A scrambled lenti-ORF-ctrl was used as a negative control. The efficacy of transduction was measured 1 week after transduction, and subsequent flow cytometry sorting of the GFPpositive cells was performed to have a pure population of GFP-positive cells on which we could perform the measurements.

Identification of Two Evolutionarily Conserved Ets Binding Sites in the FOXP3
Promoter-The functional human FOXP3 gene sequence has been previously defined (38), and the region preceding the 5Ј-untranslated region has been studied and found to be highly conserved and to contain important promoter elements such as TATA, GC, and CAAT boxes as well as binding sites for NFAT and AP-1, which are mediators of T-cell activation (55).
The Ets family of transcription factors is evolutionarily conserved and binds to a purine-rich core DNA sequence (GGAA/T), with additional flanking nucleotides often determining specificity (56). Therefore, we investigated the FOXP3 promoter region with respect to GGAA/T nucleotides using ClustalW alignment comparing human, mouse, and rat sequences.
ClustalW alignment determined that, out of several Ets binding sites located just upstream of the transcription start site in the human FOXP3 promoter region, positions Ϫ102 and Ϫ282 were conserved across species (Fig. 1). These putative Ets binding positions located just downstream of the binding sites for the transcription factors NFAT and AP-1 (Fig. 1), were of special interest.
Ets-1 and Ets-2 Bind in Vivo to the FOXP3 Promoter in Natural Tregs-Given the high expression of FOXP3 in nTregs, we investigated the in vivo binding of Ets-1 and Ets-2 to the human FOXP3 promoter, using a ChIP assay performed on nTregs. In this technique, DNA-binding proteins are covalently linked to the genomic DNA by exposure to a cross-linking agent, and then the DNA-protein complexes are isolated by immunoprecipitation (57). Following reversal cross-linking, the DNA is amplified with primers encompassing the transcription factor binding sites of interest. Thus, T cells were initially subjected to cross-linking with formaldehyde, followed by lysis and chromatin sonication. The chromatin was then immunoprecipitated with anti-Ets-1, anti-Ets-2, and anti-PU.1 antibodies and normal rabbit IgG as a background control. The resulting DNA was subjected to PCR amplification with primers flanking the region containing the two Ets binding sites. Fig. 2 shows that amplicons were generated from both anti-Ets-1 and anti-Ets-2 immunoprecipitates and not from either anti-PU.1 or control IgG. Furthermore, immunoprecipitated samples were subjected to quantitative real-time PCR (SYBR Green) using primers to specifically amplify the FOXP3 promoter region encompassing the two Ets binding sites. Anti-Ets-1 and anti-Ets-2 immunoprecipitated chromatin showed a highly significant enrichment in the FOXP3 promoter compared with IgG immunoprecipitated chromatin. Hence, these data confirm the binding of both Ets-1 and Ets-2 to the FOXP3 promoter region.
Ets-1 and Ets-2 Stimulate the FOXP3 Promoter-To ascertain whether Ets-1 and Ets-2 binding to the two Ets binding sites had a functional role, we produced two fulllength promoter constructs, a wild-type (pGL3-F-wildtype) and a mutated one (pGL3-F-Ets-mut) containing the mutation (GGAA/T3 AGAA/T) known to abrogate Ets-1 and Ets-2 binding. A transient reporter assay revealed that mutation in both Ets binding sites reduced promoter activity by 60% compared with PGL3-F-WT (Fig. 3A). In addition, cotransfection of Ets-1 and Ets-2 expression vectors with the full-length FOXP3 promoter construct resulted in an increase in promoter activity by 35 and 20%, respectively, compared with PGL3-F-WT cotransfection with pcDNA3.1, whereas cotransfection of both Ets-1 and Ets-2 constructs together with the full-length FOXP3 promoter construct resulted in an increase in promoter activity of 65% compared with PGL3-F-WT cotransfection with pcDNA3.1 (Fig. 3B). These results demonstrate that Ets-1 and Ets-2 play a positive role in the regulation of FOXP3 expression, via their binding to its promoter.

Valproate Treatment Increases FOXP3 mRNA and Protein
Levels in CD4ϩ, CD25Ϫ T Cells-To approach the epigenetic mechanisms involved in the regulation of FOXP3, we next investigated, in CD4ϩ, CD25Ϫ T cells, the impact of valproate treatment on FOXP3 expression and protein level. The effect of valproate on FOXP3 mRNA and protein levels was determined by qRT-PCR and Western blot analysis, respectively. UCB

. Ets-1 and Ets-2 bind in vivo to the FOXP3 promoter in Tregs.
ChIP assays were performed using anti-Ets-1, anti-Ets-2, anti-PU.1 antibodies, and rabbit isotype immunoglobulin G (IgG) as negative control. The binding of Ets-1, Ets-2, and PU.1 to the FOXP3 promoter was examined in regulatory T cells. Shown is the PCR for FOXP3 gene promoter from anti-Ets-1, anti-Ets-2, anti-PU.1, and IgG-immunoprecipitated chromatin. Results are representative of three independent experiments. IL-12p35 gene promoter was used as a negative control.

FIGURE 3. Effects of Ets-1 and Ets-2 on the transcriptional activity of the FOXP3 promoter-luciferase constructs.
A, Jurkat cells were transiently transfected with wild type (pGL3-F-wild type) and Ets mutated (pGL3-F-Etsmut) full-length promoter constructs to examine their luciferase activities. The data were normalized to the Renilla luciferase (pRL-TK) reporter construct. Data represent mean Ϯ S.D. of three independent experiments, each performed in triplicate. The statistical significance was determined using Student's t test (* for p Ͻ 0.05 compared with pGL3-F-WT). Transfection of pGL3-Basic vector (pGL3-BV) was used to assess the background promoter activity of the vector.pGL3-promoter vector (pGL3-PV) was used as positive control for transfection. B, the effect of Ets-1 and Ets-2 on transcriptional activation of FOXP3 promoter was confirmed by cotransfection of pGL3-F-wild type with pEts-1, or pEts-2, or combination of pEts-1 and pEts-2 expression vectors in Jurkat cells. Cotransfection of pGL3-F-wild type with pcDNA3.1 was used as a control for transfection. Data represent mean Ϯ S.D. of three independent experiments, each performed in triplicate. The statistical significance was determined using Student's t test (* for p Ͻ 0.05 compared with pGL3-F-WT cotransfected with pcDNA3.1).
CD4ϩ, CD25Ϫ T lymphocytes were cultured with and without valproate for 2, 4, 6, and 24 h. At each time point, proteins and RNA were prepared from the cells. After culture with valproate, FOXP3 expression increased in a time-dependent manner, compared with FOXP3 mRNA level in cells cultured in the absence of valproate as determined by qRT-PCR. The levels of FOXP3 mRNA increased by 2-, 2.6-, and 3-fold at, respectively, 2, 4, and 6 h after the onset of culture with valproate, compared with control untreated CD4ϩ, CD25Ϫ T cells, where no change was observed. By 24 h, the level of FOXP3 mRNA started to decrease but remained elevated slightly above the level detected in cultures without valproate (Fig. 4A). On the other hand, as shown in Fig. 4B, the level of FOXP3 protein increased after 2, 4, and 6 h culture with valproate by 1.8-, 2.2-, and 2.4-fold compared with untreated CD4ϩ, CD25Ϫ T cells. By 24 h, FOXP3 protein level started to decline but remained elevated (1.4-fold more) compared with untreated CD4ϩ, CD25Ϫ T cells.
Chromatin Studies in CD4ϩ, CD25Ϫ T Cells upon Valproate Treatment-Histone H4 hyperacetylation is a typical feature of active transcription (58). We therefore examined by ChIP the effect of HDAC inhibition on the acetylation of H4 associated with the FOXP3 promoter region as well as the binding of HDAC.
Chromatin fragments from cells cultured without and with valproate (1 mM) were immunoprecipitated with antibody to acetylated histone H4 (recognizing histone H4 acetylated at K5, K8, K12, and K16). DNA from the immunoprecipitate was isolated. From this DNA, a 200-bp fragment of the FOXP3 promoter region was amplified. Fig. 5A shows that after 6-h treatment with valproate, FOXP3 promoter DNA was associated with highly acetylated histone, compared with the same region isolated from cells cultured without valproate. By 24 h, the increase in acetylation of histone H4 associated with FOXP3 promoter region observed at 2 h was no longer detectable. Moreover, in vivo binding of HDAC to the FOXP3 core promoter was more abundant in untreated CD4ϩ, CD25Ϫ T cells compared with the same cells being treated with valproate for 6 h (Fig. 5A). On the other hand, the in vivo binding of HDAC to FOXP3 core promoter in CD4ϩ, CD25Ϫ T cells started to increase after 24-h treatment with valproate. Therefore, valproate induces a transient (24 h) H4 acetylation and inhibition of HDAC binding.
These results were further confirmed by qRT-PCR performed on anti-acetylhistone-H4 and anti-HDAC-immunoprecipitated chromatin from untreated and valproate-treated CD4ϩ, CD25Ϫ T cells. Given the above observation that valproate treatment of these cells resulted in an increase in FOXP3 expression and enhanced acetylation, we analyzed by ChIP the accessibility of the chromatin in the FOXP3 promoter area to the transcriptional machinery. An obvious increase in accessibility for Ets-1 and Ets-2 was observed in 6-h valproate-treated CD4ϩ T cells in comparison with untreated CD4ϩ T cells (Fig.  5B). Following this observation, immunoprecipitated samples were subjected to quantitative real-time PCR (SYBR Green) using primers to specifically amplify the FOXP3 promoter region encompassing the two Ets binding sites. Anti-Ets-1-and anti-Ets-2-immunoprecipitated chromatin from valproatetreated CD4ϩ, CD25Ϫ T cells showed a very significant enrichment in the FOXP3 promoter compared with IgG-immunoprecipitated chromatin. In addition, this significant enrichment observed in valproate-treated CD4ϩ, CD25Ϫ T cells was very high compared with that observed in non-treated cells. Taken together, these results demonstrate that the chromatin remodeling induced by valproate treatment results in the transient binding of Ets-1 and Ets-2, which in turn increases FOXP3 expression.
Valproate-treated UCB CD4ϩ, CD25Ϫ T Cells Transiently Adopt nTreg miRs Expression Signature-In our previous report (52), we identified the microRNA signature of human nTregs, which was composed of two down-regulated miRs (miR-31 and miR-125a), and three up-regulated miRs (miR-21, miR-181c, and miR-374). Given these data, and the fact that valproate treatment of CD4ϩ, CD25Ϫ T cells resulted in an increase in FOXP3 expression via the binding of Ets-1 and Ets-2, as shown above, we next investigated the effect of valproate on the miR profile of these cells.
This experiment revealed a striking modification of the miR profile of the treated cells. Results are shown as a -fold change in miR expression compared with the levels observed in control untreated cells. After 2 h of treatment, the miR profile of these cells began to resemble that of nTreg cells: miR-21 (Fig. 6A), miR-181c (Fig. 6B), and miR-374 (Fig. 6C) showed an increase in their expression by 1.5-fold, 2-fold, and 1.6-fold, respectively, whereas expression of miR-31 (Fig. 6D) and miR-125a (Fig. 6E) decreased by 1.6-and 1.7-fold, respectively. At later time points (4 and 6 h), miR-21 showed 2.2-and 2.6-fold increases in expression, whereas miR-181c showed 2.5-and 3.3-fold increases in expression. Concurrently, miR-374 showed 1.9-and 2.2-fold increases in its expression after 4 and 6 h of culture with valproate, respectively. By 24 h, miR-21, miR-181c, and miR-374 expression levels decreased but remained elevated compared with untreated control cells.
On the contrary, miR-31 and miR-125a relative expression levels continued to decrease, reaching their minimum after 6 h (2.2-and 2.1-fold reductions, respectively). By 24 h, the relative expression levels of both miRs started to increase, but remained below the level of control cells. In summary, CD4ϩ, CD25Ϫ UCB T cells transiently adopt an nTreg cell-like miR signature upon valproate treatment (high FOXP3 expression, low miR-31 and miR-125a expression, and high level of miR-21, miR-181c, and miR-374).

Transduction of Non-Tregs by the Lenti-ORF-FOXP3 Vector Does Not Change Their miRs Expression Profile Despite a Significant Increase in FOXP3 Protein
Level-Following GFP-positive cell sorting, FOXP3 protein level was assessed in non-Tregs after transduction by the lenti-ORFs. Fig. 7 shows that FOXP3 protein level increased by Ͼ2-fold in lenti-ORF-FOXP3-transduced cells compared with lenti-ORF-ctrl and non-transduced cells. The faint band observed in non-transduced or mock transduced cells is due to the culture conditions needed for their survival, which leads as we previously showed, to an activation of all the cells. We then assessed the miRs expression levels in lenti-ORF-transduced non-Tregs and found no differences in the levels of the five miRs constituting the signature upon FOXP3 increased levels of expression and protein (Fig. 8, A-E). This indicates that the changes in the miR expression levels observed in valproate-treated non-Tregs cells is not a consequence of the increase in expression of FOXP3. Rather, these experiments suggest that HDAC inhibition modifies the five miRs expression levels via other mechanisms and support our hypothesis that the increase of FOXP3 expression observed after valproate treatment may be due to the combina- FIGURE 5. Valproate treatment induces acetylation of H4 lysines and increases transcription factors accessibility to FOXP3 promoter. A, the acetylation status of histone H4 in the nucleosomes associated with the FOXP3 core promoter region was assessed by ChIP assay in CD4ϩ, CD25Ϫ T cells, valproate (Val)-treated CD4ϩ, CD25Ϫ T cells, and CD4ϩ, CD25ϩ regulatory T cells. Cells were lysed, and proteins were cross-linked with formaldehyde and immunoprecipitated with Ab to acetylated histone H4 (anti-acetyl H4), anti-HDAC, or control Ab (rabbit IgG). Shown is the PCR for the FOXP3 gene promoter after reversing the cross-linking. The input represents PCR amplification of the total sample, which was not subjected to any precipitation. B, ChIP assay using anti-Ets-1, anti-Ets-2, anti-PU.1, or control Ab (rabbit IgG) was carried out to study the in vivo binding of Ets-1, Ets-2, and PU.1 to FOXP3 promoter region in untreated and valproate (Val)-treated CD4ϩ, CD25Ϫ T cells. Shown is the PCR for FOXP3 gene promoter after reversing the cross-linking. Results are representative of three independent experiments. tion of two additive positive factors, i.e. the accessibility of the Ets transcription factors to their binding sites and the change in the expression of miR-21 and miR-31, which we previously showed to increase FOXP3 expression (52).

DISCUSSION
Several reports have demonstrated that HDAC inhibitors could increase FOXP3 expression (39,59), a case further illustrated in our experiments. We initially focused on mechanisms that could be responsible for this increased expression, in the UCB nTreg model we described previously (52). In that model, we had already demonstrated that two (miR-21 and miR-31) of the five miRs comprising the nTregs signature had an impact (positive and negative, respectively) on FOXP3 expression. It might be of interest to notice that the culture conditions, used both in our previous report for lentiviral transduction and in this report for all the experiments, do increase the expression of FOXP3 and modify the expression of the five miRs but preserve the ratios observed when comparing nTregs and CD4-positive CD25-negative T-lymphocytes. This explains the weak FOXP3 (mRNA and protein) positivity observed in untreated CD4ϩ, CD25Ϫ cells. One could wonder why we chose valproate, because there are more potent HDAC inhibitors available. The reason is that we wanted to take advantage of a compound  already used in humans to treat medical conditions, at concentrations comparable to the upper limit of the therapeutic ranges. We focused on the mechanisms by which HDAC inhibition by valproate could increase FOXP3 expression at the level of the FOXP3 promoter and of the two miRs quoted above that could potentially affect the level of FOXP3 mRNA and protein, in CD4-positive CD25-negative T cells. First, our experiments demonstrated a profound impact of valproate treatment on miR-21 and miR-31 expression levels that transiently became identical to their levels in nTregs. This led us to investigate the impact of valproate on the three other miRs composing the nTregs signature (125a, 181c, and 374). In our model, the level of expression of these had been shown not to impact on FOXP3 expression. Unexpectedly, we observed that these three miRs also acquired levels of expression similar to what was found in nTregs. In other words, upon valproate treatment, the CD4-positive CD25-negative cells acquired the nTregs microRNA signature. Unexpectedly, we observed that these three miRs also acquired levels of expression similar to what was found in nTregs. In other words, upon valproate treatment, the CD4-positive CD25-negative cells acquired the nTregs microRNA signature. We next asked the question whether the opening of the chromatin structure could lead to the binding of yet undescribed transcription factors that could positively affect the expression of the FOXP3 gene. The functional human FOXP3 promoter has been previously defined (38). The region preceding the 5Ј-untranslated region is highly conserved across species. This region contains important promoter elements, such as TATA, GC, and CAAT boxes, as well as binding sites for NFAT and AP-1, which are mediators of T-cell activation (55). This region was found also to contain several potential binding sites for the Ets family transcription factors, a purine-rich core sequence with additional flanking nucleotides determining specificity. Therefore we investigated the FOXP3 promoter region with respect to GGAA/T nucleotides by comparing human, mouse, and rat sequences. As described under "Results," we observed that, out of several Ets binding sites located upstream of the transcription start site in the human FOXP3 promoter region, two positions, Ϫ102 and Ϫ282, were well conserved between species. Because these two Ets binding sites are located just downstream of the binding sites for the transcription factors NFAT and AP-1, we per- , and miR-374 (E) expression levels were analyzed by qRT-PCR following lenti-ORF transduction. Shown are the relative levels (mean Ϯ S.D.) of three independent experiments performed on three donors, each done in triplicate. There were no statistically significant differences between FOXP3transduced and control CD4ϩ, CD25Ϫ cells for any of the five miRs, as determined by Student's t test.

Valproate Confers on CD4؉, CD25؊ Cells a Treg Profile
formed a series of experiments in nTregs, using the ChIP technique, showing that these two factors bound to these sites: Ets-1 and Ets-2. Next, using reporter constructs containing the fulllength FOXP3 promoter, either wild-type or mutated in these two binding sites, we could then demonstrate, in transfection experiments, that the binding of these two Ets transcription factors up-regulated FOXP3 expression. Next, ChIP experiments were performed in non-Tregs with and without valproate treatment. We could show that the binding of these two factors occurred only when the chromatin was open. This was further confirmed by the demonstration that valproate treatment of non-Tregs strongly increased the acetylation status of histone H4 associated to the FOXP3 promoter region, thereby allowing the access of the two Ets factors to their FOXP3 binding sites. Taken together, these results demonstrate that one way by which HDAC inhibitors regulate FOXP3 expression is by modulating the binding of Ets-1 and Ets-2 transcription factors to the promoter region.
The next question we wanted to ask was whether the two phenomena (the increase in FOXP3 expression and the acquisition by CD4-positive, CD25-negative UCB T cells of a Treg miR profile) were intricately linked or not. To answer this question, we assessed the miR expression levels in non-transduced, lenti-ORF-ctrl-and lenti-ORF-FOXP3-transduced non-Tregs. No difference in the level of the five miRs, constituting the signature, in lenti-ORF-FOXP3-transduced non-Tregs was found, compared with the controls, despite FOXP3 increased levels of mRNA and protein by Ͼ2-fold. This indicates that the change of the miR expression levels observed in the valproatetreated non-Tregs is not a consequence of the increase in expression of FOXP3. Rather, these experiments suggest that HDAC inhibition modifies the expression levels of the five miRs via other mechanisms and support our hypothesis that the increase of FOXP3 expression observed after valproate treatment may be due to the combination of two additive positive factors, i.e. the accessibility of the Ets transcription factors to their binding sites and the change in the expression of miR-21 and miR-31, which we previously showed to increase FOXP3 expression (52).
To summarize these observations, one could say that valproate treatment (at concentration similar to the upper limit of the therapeutic window in humans) confers transiently to non-Treg CD4-positive T cells a molecular signature (microRNAs and FOXP3 expression) similar to the one of natural Tregs.
A yet unanswered mechanistic question is how valproate modifies the expression of the five miRs we have tested, in opposite directions. Indeed, HDAC inhibition could account for an increased expression of some miRs, but the mechanism leading to decreased expression has to be indirect. Anyway, it is not due to the increased FOXP3 expression level. Answering this question will require further investigations.
Nevertheless, our findings provide novel insights on the mechanisms by which epigenetic manipulations of non-Treg CD4-positive cells result in the modulation of FOXP3 expression and modification of their miR signature. They also open new ways of research that could lead to novel approaches in the treatment of human diseases linked to a deficit of Tregs.