OX40 Gene Expression Is Up-Regulated by Chromatin Remodeling in Its Promoter Region Containing Sp1/Sp3, YY1, and NF-κB Binding Sites¹

OX40 (CD134) is a member of the TNFR superfamily (TNFRSF4) (1) that was originally identified as a rat T cell activation marker (2). On one hand, OX40 expression was determined to be restricted to T cells and up-regulated upon signaling through the TCR (3). On the other hand, OX40 ligand (OX40L)³ is expressed on a number of cell types, including professional APCs (dentritic cells and macrophages) (1, 4). OX40/OX40L interaction plays an important role in CD4⁺ Th cell responses (1, 4) and the generation of memory T cells (5, 6). Although OX40 can contribute to CD8⁺ T cell responses, many studies have suggested a preferential role for this receptor in CD4⁺ Th2 responses (1, 4). Indeed, OX40 signaling, in conjunction with TCR and CD28 signaling, induces NFATc1 accumulation in the nucleus to drive initial IL-4 transcription (7). Both OX40 (8)- and OX40L (9)-deficient mice exhibit defects in CD4⁺ T cell proliferative responses, and in transgenic mice overexpressing OX40L in dendritic cells, CD4⁺ T cells accumulate in the B cell follicles (10). Furthermore, OX40 is preferentially expressed on activated CD4⁺ T cells under certain conditions (11). Signaling through OX40 can promote survival signals in effector T cells (12, 13) and can also activate NF-κB through a TNFR-associated factor 2- and 5-mediated pathway (14).

OX40 is expressed at high levels on CD4⁺CD25⁺ regulatory T (Treg) cells (15, 16), suggesting that it might play an important role in these cells. Indeed, signaling through this receptor can inhibit suppressive activity of these cells (17), paralleling the function of glucocorticoid-induced TNFR (TNFRSF18) (15, 18). This molecule, like OX40, is also mainly expressed on T cells, and functions as a costimulatory receptor on effecter T cells (19). Why CD4⁺CD25⁺ Treg cells have these two receptors with similar functions, and how these cells can regulate high-level expression of these two receptors are currently unknown. Thus, although the role of OX40 in T cell-mediated immune responses has been well documented, and even studied in autoimmunity and immune responses to cancer (4), regulation of its gene expression in activated effector T cells, as well as in CD4⁺CD25⁺ Treg cells, is poorly understood.

In the present study, we show that OX40 mRNA expression is up-regulated in the T cell line EL4 by activation with anti-CD3 Ab (same as that in primary T cells). This suggests that we can use this cell line as a model system to study OX40 gene expression. Using the EL4 line, we show that basal OX40 promoter activity is regulated by the constitutively expressed transcription factors Sp1/Sp3 and YY1. Significantly, we demonstrate drastic activation-associated chromatin remodeling in the OX40 basal promoter region. OX40 gene expression seems to be regulated by chromatin remodeling upon activation, and NF-κB might be involved in initiation of this chromatin remodeling. Importantly, in CD4⁺CD25⁺ Treg cells (without any in vitro activation), chromatin structure in the OX40 basal promoter region is relaxed by histone acetylation similar to that in activated CD4⁺CD25⁻ T cells.

EL4 cells were cultured in IMDM+GlutaMax (Invitrogen Life Technologies) with 5% FBS, and primary T cells were cultured in RPMI 1640 with l-glutamine (Lonza) with 10% FBS. Penicillin (10 U/ml)-streptomycin (10 μg/ml) (Invitrogen Life Technologies) was also added into these media. Total primary T cells (for RT-PCR and luciferase assays) and CD4⁺ T cells (for RT-PCR) were isolated from mouse spleen cells using a pan-T cell isolation kit (Miltenyi Biotec) and a CD4⁺ T cell isolation kit (Miltenyi Biotec), respectively. For the chromatin immunoprecipitation (ChIP) assay, CD4⁺CD25⁻ and CD4⁺CD25⁺ Treg cells were purified by a cell sorter using allophycocyanin-anti-CD4 Ab (L3T4; BD Pharmingen) and bitotin-anti-CD25 Ab (7D4; BD Pharmingen) with streptavidin-PE (BD Pharmingen). Purity of CD4⁺CD25⁻ and CD4⁺CD25⁺ T cells were 99 and 98%, respectively. If required, cells were activated by culturing with anti-CD3 Ab (KT3) precoated (with 5 μg/ml Ab in PBS) plates.

OX40 mRNA levels were analyzed by RT-PCR as described previously (20). Briefly, cDNA was prepared using an oligo(dT) primer and 1 μg of total RNA from EL4 cells, primary total T cells, and CD4⁺ T cells. To perform semiquantitative RT-PCR, PCR was stopped at different cycle numbers (OX40: 21, 24, 27, and 30 cycles; 18S rRNA: 13, 16, and 19 cycles). PCR products were analyzed by agarose gel with ethidium bromide staining. Quantitative PCR was performed by real-time PCR using TaqMan 7900HT (Applied Biosystems) with Power SYBR Green PCR Master Mix (Applied Biosystems). The PCR primers used were as follows: OX40 sense, GTAGACCAGGCACCCAACC; OX40 antisense, GGCCAGACTGTGGTGGATTGG; 18S rRNA sense, CTTAGAGGGACAAGTGGCG; and 18S rRNA antisense, ACGCTGAGCCAGTCAGTGTA.

The OX40 transcription start site was determined by RACE (21) with minor modifications as described previously (20). cDNA was prepared using an antisense oligo primer (GTTGCACTGTGTACACTGCTTG) and total RNA from nonactivated and activated CD4⁺CD25⁻ T cells, CD4⁺CD25⁺ Treg cells, and activated EL4 cells. After adding a poly(G) tail at the resulting cDNA end, OX40 cDNA were amplified using a poly(C) primer and an antisense primer (ATTGTAGAAGCCAGTCTCACATG). To concentrate OX40 cDNA fragments, the resulting PCR products were amplified again using the second antisense primer (TGGCACTCACGACAGCACTTGTG). The resulting PCR products were cloned and DNA sequences were determined. To avoid PCR artifacts as a result of this procedure, seven independent PCR products were analyzed.

OX40 promoter fragments (11 OX40 promoter deletion mutants, Sp1 site knockout (KO), YY1 site KO, and NF-κB site KO) were cloned into the pGL4 Basic Vector (Promega). All DNA sequences of inserted fragments were determined to remove deformed fragments generated by PCR errors. For the luciferase assay, 5 × 10⁶ EL4 cells or 3 × 10⁶ primary total T cells were transfected with 5 μg (for EL4) or 3 μg (for primary T cells) of luciferase reporter plasmids and 1 μg (or 0.5 μg for activated condition) of pRL-CMV as an internal control plasmid, and cultured in six-well plates. For EL4 and primary T cell transfection, Gene Pulser Xcell (Bio-Rad) and Nucleofector with mouse T cell kit (Amaxa) were used, respectively. To activate cells, transfected cells were transferred into anti-CD3 Ab (KT3) precoated (with 5 μg/ml Ab in PBS) six-well plates at 4 h posttransfection, and cultured for additional 44 h. Cells were harvested, and promoter activities were analyzed by Dual-Luciferase Reporter Assay System (Promega). These assays were repeated at least three times, and firefly luciferase (pGL4) activities were normalized to Renilla luciferase (pRL-CMV, internal control) activities.

Nuclear extracts were prepared using nonactivated EL4 and anti-CD3 activated EL4 and total primary T cells (for 24 h) (for Fig. 3,B) (for 2 h) (for Fig. 5 C). Harvested cells were homogenized in hypotonic lysis buffer (10 mM HEPES (pH 7.9), 1.5 mM MgCl₂, 10 mM KCl, and 1 mM DTT with proteinase inhibitor mixture (Sigma-Aldrich)), and then nuclei were collected by centrifugation. Nuclear proteins were extracted with extraction buffer (20 mM HEPES (pH 7.9), 1.5 mM MgCl₂, 0.42 M NaCl, 0.2 mM EDTA, 25% glycerol, 1 mM DTT with protease inhibitor mixture (Sigma-Aldrich)). EMSA was performed with 2 μg of nuclear extract in 20 μl of EMSA reaction buffer (2 μg of poly(dI-dC)poly(dI-dC), 20 mM HEPES (pH 7.9), 1 mM MgCl₂, 40 mM KCl, 0.1 mM EDTA, 1 mM DTT, and 12% glycerol). To perform the competition assay, a 100-fold excess of unlabeled oligo competitor primer was added to the EMSA reaction mixture. Samples were analyzed on a polyacrylamide gel containing 89 mM Tris borate and 20 mM EDTA. For the supershift assay, nuclear extracts in EMSA reaction buffer were incubated with anti-Sp1 (Santa Cruz; PEP2), anti-Sp3 (Santa Cruz; D-20), anti-YY1 (Santa Cruz; H-10), NF-κB p65 (Santa Cruz; C-20), and NF-κB p50 (Santa Cruz; D-17) Abs for 15 min, at which time probes were then added. Probe and competitor sequences (sense strand) used were as follows: P1, ACGCCTGTGCCAAATACACAGGAACACGTT; P2, GGAACACGTTCACATACCTTCTTGCCTGTC; P3, CTTGCCTGTCCGCCTACTCTTCTTGCCCCA; P4, TCTTGCCCCACCTCCATAGTTCTTATAGCC; P5, ACCTCCATAGTTCTTATAGCCACAC; P6, TTATAGCCACACCCTGCAAGGAAAA; P7, CCTGCAAGGAAAAACCCCAGACTCC; Sp1A KO, TCTGAATTCACCTCCATAGTTCTTATAGCC; YY1 KO, TCTTGCCCCACCTCCGAATTCCTTATAGCC; Sp1A/YY1 KO, TCTTGCCCCACGAATTCAGTTCTTATAGCC; Sp1B KO, TTATAGCCACAGAATTCAAGGAAAA; and NF-κB KO, CCTGCAAGGACTCGAGCCAGACTCC. DNA sequences of IL-10/Sp1 (20), CD40/Sp1 (22), and CD40/NF-κB (NF-κBA site) (22) competitors were shown previously.

ChIP assay was performed using EL4 cells, CD4⁺CD25⁻ T cells, and CD4⁺CD25⁺ Treg cells. These cells were fixed in fixation solution (1% formaldehyde, 4.5 mM HEPES (pH 8.0), 9 mM NaCl, 0.09 mM EDTA, and 0.045 mM EGTA) for 10 min at room temperature. The reaction was stopped by adding glycine solution (final concentration, 0.1 M). Fixed cells were then washed twice with cold PBS, and placed in lysis buffer (1% SDS, 10 mM EDTA, and 50 mM Tris-HCl (pH 8.0) with protease inhibitor mixture (Sigma-Aldrich)). Chromatin was sheared by sonication on ice, and the insoluble fraction was removed by centrifugation. This lysate was precleared with control Ig plus salmon sperm DNA-protein A agarose (Upstate) and Abs were added and incubated at 4°C overnight. Immunocomplexes were collected with salmon sperm DNA-protein A agarose, and washed twice with low salt buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.0), and 150 mM NaCl), twice with high salt buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.0), and 500 mM NaCl), LiCl wash buffer (0.25 M LiCl, 1% IGEPAL-CA630, 1% deoxycholic acid, 1 mM EDTA, 10 mM Tris-HCl (pH 8.0)), and TE (10 mM Tris-Cl (pH 7.5) and 1 mM EDTA). Precipitated chromatin fragments were eluted with elution buffer (1% SDS, 0.1 M NaHCO₃), after adding NaCl (final concentration, 0.2 M), and samples were incubated at 65°C overnight. After treatment with proteinase K, DNA was extracted using PCR DNA purification kit (Qiagen). Quantitative real-time PCR analysis with Power SYBR Green PCR Master Mix (Applied Biosystems) was performed to determine the percentage of OX40 basal promoter precipitated with Abs against input DNA. PCR primers used to amplify the OX40 basal promoter and Abs were as follows: OX40 promoter sense, TACGCCTGTGCCAAATACAC; OX40 promoter antisense, GCTTTCTGCCTTCACAGGAG; anti-p105/p50-ChIP grade (Abcam); anti-acetyl histone H4 polyclonal Ab (Upstate); and rabbit IgG (Santa Cruz Biotechnology).

Because up-regulation of OX40 in activated T cells is key to physiology of this receptor, we investigated how OX40 gene expression is regulated. To determine whether OX40 expression is controlled by transcriptional and/or posttranscriptional regulation in activated T cells, RT-PCR for OX40 mRNA was performed using RNA from the T cell line EL4 (Fig. 1), primary total T cells (A), and CD4⁺ T cells (B). To perform semiquantitative RT-PCR, the PCR was stopped at different cycle numbers (Fig. 1,A). OX40 mRNA was detected in nonactivated total primary T cells and EL4 cells, and up-regulated in cells activated with anti-CD3 Ab (Fig. 1,A). PCR products from OX40 mRNA were observed in earlier PCR cycle numbers for primary T cells (both activated and nonactivated), compared with EL4 cells, indicating that the expression level of OX40 mRNA is higher in the former. To analyze the induction of OX40 mRNA expression during activation with anti-CD3 Ab, real-time PCR was performed using RNA from EL4 and CD4⁺ T cells. OX40 mRNA levels were increased 18- and 12-fold by activation in EL4 and CD4⁺ T cells, respectively (Fig. 1 B). These findings suggest that OX40 expression is at least partially controlled by transcriptional and/or posttranscriptional (e.g., stabilization of OX40 mRNA in activated cells) regulation, and that we can use EL4 cells as a model system to study the regulation of OX40 expression. In the 3′-untranslated region (UTR) of OX40 mRNA, we could not identify a typical mRNA destabilization signal (AUUUA+AU-rich sequence) (23, 24) (we have also previously detected in IL-10 mRNA (25)), we chose to focus on the transcriptional regulation of OX40 gene expression.

As a first step to investigate the transcriptional regulation of OX40 gene expression, we examined the OX40 promoter activity. To determine the location of the OX40 promoter, we analyzed the transcriptional start site by 5′-RACE, using RNA from CD4⁺ T cells, CD4⁺CD25⁺ Treg cells, and EL4 cells. In total, we analyzed 71 5′-RACE clones to determine the major mRNA start site, and to avoid artifacts resulting from the PCR procedure, seven independent PCR products were used for this assay (two PCR products from nonactivated CD4⁺ T cells, two PCR products from activated CD4⁺ T cells, two PCR products from CD4⁺CD25⁺ Treg cells and a PCR product from activated EL4 cells). The same site was mapped as a major mRNA start site in all reactions (55 clones of total 71 clones) at 8 bp upstream of the first ATG of OX40 mRNA, and this mRNA start site was defined as position +1. The same major transcription start site was determined in all T cells tested, indicating that the OX40 promoter could be located upstream of this start site in nonactivated and activated cells. No TATA box sequence was found in the region 30 bp upstream of the major transcription start site, suggesting that OX40 gene expression is regulated by a TATA-less promoter. A 1.97-kb promoter fragment (Pro 1) (Fig. 2) containing the major mRNA start site was PCR-amplified from mouse genomic DNA, and 10 deletion mutants of this promoter (all containing the major mRNA start site) (Pro 2 to Pro 11) (Fig. 2) were generated by PCR and cloned into the pGL4 Basic Vector (Promega) (Fig. 2). The resulting luciferase reporter plasmids were transfected into EL4 cells, and luciferase activities were compared with that of the negative control plasmid (pGL4 Basic Vector, no promoter). A significant reduction of promoter activity resulted from deletion of 69 bp (between Pro 9 and Pro 10) and 30 bp (between Pro 10 and Pro 11) regions in both nonactivated EL4 cells and anti-CD3 Ab-activated cells. This finding suggested that OX40 basal promoter activity is located between −125 (5′ end of Pro 9 fragment) and +4 (3′ end of all promoter fragments), and constitutively expressed transcription factors might be binding to this basal promoter region. This OX40 basal promoter activity was also analyzed in activated primary T cells by a luciferase assay using selected promoter constructs, and regulatory regions were mapped in the identical positions (Fig. 2 C).

Having concluded via OX40 promoter deletion mutants that regulatory elements for OX40 promoter activity are located between position −125 (5′ end of promoter 9 fragment) and −26 (5′ end of promoter 11 fragment), we next sought to identify transcription factors binding to this region. EMSA was performed using seven oligo probes containing sequences in this region (Fig. 3). Four slowly migrating complexes were detected by EMSA using the P4 probe, with both nuclear extracts from nonactivated and activated EL4 cells (Fig. 3,B). All four complexes (C1 to C4) with ³²P-labeled P4 probe disappeared with a 100-fold excess of unlabeled P4 competitor, but not with P2 (Fig. 4,A), indicating sequence-specific complex formation. To determine the transcription factor binding regions in this P4 probe, EMSA was performed using seven mutant P4 oligo probes (results using three mutant probes are shown in Fig. 6, B and C, but results using other mutants are not shown). This result suggests that these factors bind within a TGCCACCTCC sequence. The potential Sp1 family binding sequence was identified in this sequence by the transcription factor database (TESS). These results suggest that complexes C1, C2, and C3 could form with the transcription factor Sp1 family on a GCCCCACCTC sequence (position −70 to −61). We therefore performed competition assays using P4 probe with CD40 (22) and IL-10 (20) Sp1 oligo competitors that we used previously for analysis of CD40 and IL-10 promoter activity. Indeed, these competitors inhibited C1, C2, and C3 complex formation with ³²P-labeled P4, but not C4 formation (Fig. 4,A). Complex formation with Sp1 (C1) and Sp3 (C2 and C3) was also confirmed by supershift EMSA using anti-Sp1 and anti-Sp3 Abs (Fig. 5 A). Two complexes with Sp3 were presumably formed with two different-sized Sp3 molecules as described previously (26).

EMSA using mutant P4 probes (data not shown) also suggests that the factor in complex C4 binds within a CCACCTCCATAG sequence. The transcription factor database suggests that complex C4 might form with transcription factor YY1 on ACCTCCATA (position −65 to −58) in the identified sequence, which was confirmed by supershift EMSA using an anti-YY1 Ab (Fig. 5 A).

Four slowly migrating complexes (C5 to C8) were also detected by EMSA using P6 probe, with nuclear extracts from nonactivated and activated EL4 cells (Figs. 3,B and 4,B). C5, C6, and C7 complex formation with ³²P-labeled P6 was inhibited with a 100-fold excess of unlabeled P6, but not with P5 (Fig. 4,B), indicating sequence-specific complex formation. However, C8 complex formation was not inhibited by either unlabeled P6 competitor (containing exactly the same sequence as in P6 probe) or P5, suggesting that the factor in complex C8 might recognize the 5′ end sequence with phosphate (³²P) in the P6 probe. Indeed, a 2-bp mutation at the 5′ end of this probe inhibited C8 complex formation (data not shown). These results suggest that C8 complex formation is an artifact due to this procedure using oligo probes. EMSA using mutant probes (data not shown) suggests that these factors bind within a TAGCCACACCCT sequence; the transcription factor database also suggests that complexes C5, C6, and C7 might form with transcription factor Sp1 family on GCCACACCCT sequence (position −48 to −39) in the identified sequence; and this was confirmed by competition assay (Fig. 4,B) and supershift EMSA using anti-Sp1 and anti-Sp3 Abs (Fig. 5,B). Two Sp1/Sp3 binding regions are present in the OX40 promoter, which we have named Sp1A (the Sp1 site in P4) and Sp1B (the Sp1 site in P6), respectively (Fig. 6,A). These transcription factors binding to the OX40 promoter was confirmed by competition assays using a nuclear extract from activated primary T cells (Fig. 5 D).

To examine the contribution of Sp1, Sp3, and YY1 to OX40 promoter activity, each transcription factor binding site was mutated (Fig. 6) in the OX40 promoter constructs Pro 2 (1.35 kb) (Figs. 2 and 6) and Pro 9 (0.13 kb) (Figs. 2 and 6), and luciferase reporter assays were performed. The structures of the resulting plasmids are illustrated in Fig. 6,A, and mutated sequences are shown with wild-type sequences in Fig. 6,B. No binding of Sp1/Sp3 or YY1 to the mutated sequences was detected by EMSA (Fig. 6 C). A significant reduction of promoter activity with the mutation of each of the transcription binding regions was observed, in both nonactivated and activated EL4 cells. No promoter activity (or only very weak promoter activity in activated cells) was detected with the triple mutation (Sp1A, YY1, and Sp1B) in this promoter, indicating OX40 basal promoter activity is regulated by constitutively expressed transcription factors, Sp1/Sp3 and YY1.

Although luciferase activity generated using Pro 1 to Pro 9 reporter constructs in activated cells was ∼1.5- to 2-fold higher than that in nonactivated cells (Fig. 2), we could not confirm the presence of a CD3 activation response element in this promoter region for the following reasons: 1) taking the variability of luciferase assays into consideration, a 1.5-fold difference might be not significant; 2) the transcriptional environments in between nonactivated and activated cells are not the same (e.g., amount of polymerase), and any observed enhancement of promoter activity could simply be a consequence of the different environments. Thus, CD3 response elements must be identified with other assay systems, such as EMSA and luciferase assay using point mutated promoter constructs.

By EMSA with probe P7 (Fig. 3,B), we detected a transcription factor bound to this probe in nuclear extracts from activated EL4 cells, but not nonactivated cells. The transcription factor database suggested that NFAT might bind to the AGGAAAAA (−35 to −28) sequence in the P7 probe. Indeed, this sequence is similar to NFAT consensus sequences in the IL-2 promoter (GAGGAAAA). We therefore performed a supershift assay using anti-NFAT Ab; however, NFAT binding to the promoter region was not detected. Because the structural similarity and heterodimer formation of NFAT and NF-κB p50 has been recently suggested (27, 28, 29, 30, 31), we also performed a supershift assay using anti-NF-κB p50, p52, p65, rel-B, and c-rel Abs. The complex formation was further shifted with both p50 and p65 Abs (Fig. 5,C), indicating that NF-κB p50 and p65 bind to the OX40 basal promoter region. NF-κB binding was also confirmed by EMSA and a competition assay using a nuclear extract from activated primary T cells (Fig. 5,D). However, we could not detect any CD3 activation response elements in the OX40 promoter region using the deletion mutants by luciferase reporter assay (Fig. 2). We therefore mutated the NF-κB binding region in the OX40 Pro 9 construct and the luciferase activity generated using the resulting plasmid was compared with that using wild-type Pro 9 in nonactivated and activated EL4 cells (Fig. 7). The NF-κB binding sequence and the mutated sequence are shown in Fig. 7,A, and the lack of NF-κB binding to the mutant sequence was also confirmed by EMSA (Fig. 7,B). OX40 promoter activity was reduced to 60% by mutation in activated cells (Fig. 7,C), but this promoter activity with the mutated sequence was also slightly reduced in nonactivated cells (90%). Considering the variability of the luciferase assay and the result in Fig. 2 (no CD3 response elements were detected by luciferase assay with a series of deletion mutants of the OX40 promoter), we cannot conclude any contribution of NF-κB in the direct enhancement of OX40 promoter activity. The up-regulation of OX40 mRNA observed in activated EL4 cells and primary T cells (Fig. 1) is also unexplained.

We hypothesized that NF-κB may contribute to chromatin remodeling in the basal promoter in vivo, rather than direct enhancement of OX40 promoter activity. OX40 gene expression may be up-regulated by chromatin remodeling in activated T cells. To investigate this possibility, induction of OX40 mRNA expression (by RT-PCR) and histone H4 acetylation in the basal OX40 promoter region containing Sp1, YY1, and NF-κB binding regions (by ChIP assay) were monitored at different time points after activation with anti-CD3 Ab in EL4 cells (Fig. 8,A). Induction of histone H4 acetylation was increased gradually after activation, and induction of OX40 mRNA expression was found to reflect the ratio of histone H4 acetylation in the basal promoter region (Fig. 8 A).

OX40 gene expression seems to be up-regulated by chromatin remodeling in the OX40 basal promoter region, and NF-κB may contribute to this chromatin remodeling event. To investigate this possibility, we performed ChIP assay using anti-NF-κB p50 Ab in EL4 and CD4⁺CD25⁻ T cells. Binding of NF-κB p50 to the basal OX40 promoter region was detected in activated EL4 and CD4⁺CD25⁻ T cells in vivo (Fig. 8,B). We have also detected binding of NF-κB p50 and p65 to the basal OX40 promoter region by EMSA (Fig. 5, C and D). Taken together, these findings indicate that NF-κB might contribute to the up-regulation of OX40 gene expression through chromatin remodeling in activated T cells.

To examine whether histone H4 is acetylated in the basal OX40 promoter region in activated CD4⁺CD25⁻ T cells and CD4⁺CD25⁺ Treg cells (which expresses OX40 at high levels), ChIP assay was performed. Acetylation of histone H4 in the basal OX40 promoter region in CD4⁺CD25⁻ T cells is increased by activation, and interestingly, in CD4⁺CD25⁺ Treg cells, histone H4 molecules in this promoter region are highly acetylated. We isolated these Treg cells from mouse spleen cells and immediately used them for ChIP assay, indicating that histone H4 molecules in the OX40 promoter region are highly acetylated in these cells without activation in vitro. The chromatin structure of the OX40 promoter region is configured to express OX40 at high levels in CD4⁺CD25⁺ Treg cells.

In this study, we have studied the up-regulation of OX40 gene expression in T cells, on the premise that up-regulation of OX40 on T cell surfaces is critical for its physiological function. We have focused on transcriptional control of the OX40 gene, and in this first analysis of the OX40 promoter, we report that OX40 basal promoter activity is regulated by constitutively expressed transcription factors, Sp1/Sp3 and YY1. However, given that OX40 mRNA levels are low in nonactivated cells, the OX40 promoter seems to function fully with activation-induced transcription factors in activated T cells. Although we identified NF-κB binding to the OX40 promoter region, there was only a 1.5-fold direct enhancement of OX40 promoter activity by this transcription factor (as determined by a luciferase reporter assay using KO promoter constructs (Fig. 7)). Thus, NF-κB-driven enhancement of OX40 promoter activity does not on its own account for the large up-regulation of OX40 mRNA expression in activated cells. We considered three possibilities: 1) OX40 gene expression may be controlled by posttranscriptional mechanisms (i.e., OX40 mRNA might be unstable in resting T cells and stabilized via activation); 2) OX40 gene expression may be up-regulated by chromatin remodeling in activated cells; and/or 3) OX40 gene expression may be up-regulated by an activation-specific enhancer. Several considerations steered us away from the first possibility, namely, posttranscriptional control. We have previously shown that IL-10 mRNA levels in EL4 cells are regulated by posttranscriptional regulation, through AUUUA+AU-rich mRNA destabilization signals located in the 3′-UTR (25). Although there is an AUUUUA sequence in the 3′-UTR of OX40 mRNA, it is not accompanied by AU-rich sequences, and therefore, OX40 mRNA seems to lack the mRNA destabilizing sequence configuration of IL-10 mRNA (25) and other mRNAs (23, 24). Additionally, no potential micro-RNA binding regions can be identified in OX40 mRNA, using the micro-RNA database (PicTar).

Consequently, we turned our attention to the second possibility, namely, chromatin remodeling of OX40 gene. Using a ChIP assay, we strikingly found that histone H4 molecules in this region are highly acetylated upon T cell activation, and that NF-κB binds to the OX40 basal promoter region in vivo. Providing a potential link between these two findings, a recent report describes that NF-κB can recruit a number of different coactivators, including p300/CBP, p/CAF, and SRC-1 (32, 33), which have histone acetyltransferase (HAT) activity. Taken together, these findings suggest that OX40 gene expression is up-regulated by chromatin remodeling in activated T cells, and NF-κB might be involved in the initiation of this chromatin remodeling.

Although we cannot rule out involvement of an activation-specific enhancer in regulating OX40 gene expression, our OX40 promoter chromatin remodeling findings prompt us to posit the following sequence of events accounting for OX40 up-regulation: 1) in resting T cells, chromatin in the OX40 promoter region is in the closed form, albeit some Sp1, Sp3, and YY1 can access this promoter and yield small amounts of OX40 mRNA; 2) NF-κB is nuclear translocalized by activation, binds to the OX40 basal promoter, and recruits HAT and subsequently acetylate histone molecules to this promoter region; and 3) once OX40 promoter chromatin structure is relaxed by histone acetylation, Sp1, Sp3, and YY1 transcriptional factors can more easily access the binding regions, leading to the up-regulation of OX40 mRNA production.

Although this study has focused on the up-regulation of OX40 gene expression, subsequent down-regulation is also of interest. OX40 up-regulation peaks at 48 h, and then declines again (3). The chromatin remodeling mechanism we have proposed in this study to account for OX40 up-regulation might be extended to also explain its down-regulation as well. Specifically, it is possible that down-regulation is mediated by deacetylation of histone molecules in the OX40 promoter region. Not only can NF-κB recruit HATs, but NF-κB p50 can additionally recruit histone deacetylase 1 (HDAC1) (34). Thus, in the late activation stage, NF-κB p50 might recruit HDACs and deacetylate histone molecules to the OX40 promoter region, closing the chromatin structure and thereby down-modulating OX40 gene expression.

In this study, we focused on regulation of OX40 expression by signaling through the TCR/CD3 complex. Enhancement of OX40 expression was also observed with costimulatory signals through CD28 (12, 35). Involvement of activated STAT6 on OX40 gene expression by CD28 signaling was excluded by a study using STAT6-deficient mice (36). CD28 signaling might up-regulate OX40 gene expression by indirect pathways. Indeed, we could not identify a CD28 response element by luciferase assay in the OX40 promoter region (data not shown).

There is an interesting possibility that the constitutive Sp1 and YY1 transcription factors might contribute to OX40 up-regulation in more than one way. Beyond directly regulating OX40 basal promoter activity, they could potentially contribute on the chromatin remodeling side as well. Recently, Sp1 interactions with HAT (p300) and HDAC1 (37, 38) have been reported, as well as ability of YY1 to recruit HAT (39) and HDACs (40, 41). Thus, histone acetylation in the OX40 promoter region might be regulated by cooperation between HATs and HDACs that are associated with both induced (NF-κB) and constitutively expressed (Sp1 and YY1) transcription factors.

YY1 can reportedly regulate transcription as both a positive and a negative regulator (42). Because OX40 promoter activity was reduced to about one-half by mutation of the YY1 site in this promoter region, YY1 seems to function largely as a positive regulator in the case of OX40 gene expression. This YY1 site actually overlaps the Sp1A site (Fig. 6 B), which is of interest given that interaction between Sp1 and YY1 has been suggested in the case of other genes (43, 44). This could be the case for OX40 gene expression as well.

Approximately 5% of CD4⁺ T cells are CD4⁺CD25⁺ Treg cells that express OX40 at high levels, and signaling through this receptor can neutralize the suppressive activity of these Treg cells (17). Significantly, we have found that histone molecules in this promoter region are highly acetylated in CD4⁺CD25⁺ Treg cells (absence of activation in vitro). Thus, chromatin structure of OX40 promoter might be always in the relaxed form in Treg cells. Building upon our description of chromatin remodeling of the OX40 promoter region in the course of T cell activation, we further posit that the differentiation of Treg cells from precursors (presumably driven by some self-Ag) may also be associated with relaxing of chromatin structure in the OX40 promoter region. This region would then be kept open in mature CD4⁺CD25⁺ Treg cells by some unknown molecular mechanism. Detailed study of chromatin remodeling in the OX40 promoter region in CD4⁺CD25⁺ Treg cells could thus potentially contribute to an understanding of how transcriptional regulation is controlled in CD4⁺CD25⁺ Treg cells and how this category of cells develops.

The authors have no financial conflict of interest.