Anti-Leukemia Activity of MS-275 Histone Deacetylase Inhibitor Implicates 4-1BBL/4-1BB Immunomodulatory Functions

Histone deacetylase inhibitors (HDACi) have demonstrated promising therapeutic potential in clinical trials for hematological malignancies. HDACi, such as SAHA/Vorinostat, Trichostatin A, and MS-275 were found to induce apoptosis of leukemic blasts through activation of the death receptor pathway and transcriptional induction of the Tumor Necrosis Factor (TNF)-related pro-apoptotic family members, TRAIL and FasL. The impact of HDACi on TNF-related costimulatory molecules such as 4-1BB ligand (4-1BBL/TNFSF9) is however not known. Following exposure to SAHA/Vorinostat, Trichostatin A, and MS-275, transcript levels were determined by real time PCR in Jurkat, Raji and U937 cells. Treatment with HDACi up-regulated TNFSF9 gene expression in the three leukemia cell lines, yet to different extend and with distinct kinetics, which did not require de novo protein synthesis and was not associated with DNAse I hypersensitive chromatin remodeling. Transcriptional activity of TNFSF9 promoter-luciferase constructs was induced up to 12 fold by HDACi, and implication of Sp1/Sp3 transcription factors binding to functional GC-box elements was evidenced by reporter gene assays, site-directed mutagenesis, and electrophoretic mobility shift assays. Functionality of modulated target genes was assessed in allogeneic mixed leukocyte reaction experiments. MS-275- and to a lesser extent Trichostatin A- and SAHA-treated Raji cells significantly up regulated T lymphocytes proliferation which was reduced by about 50% by a 4-1BB blocking recombinant protein, while MS-275- but neither Trichostatin A- nor SAHA-treated cells up-regulated IFNγ secretion by T lymphocytes. Our results identify 4-1BBL/4-1BB as a downstream target of HDACi, especially of MS-275 anti-leukemia action in vitro. Thus, HDACi such as MS-275 displaying dual TNF-dependent proapoptotic and costimulatory activities might be favored for inclusion in HDACi-based anti-cancer therapeutic strategies.


Introduction
Members of the Tumor Necrosis Factor ligands and receptors Superfamily (TNFSF) are both secreted and membrane-bound factors that regulate proliferation, activation, differentiation, maturation and survival or programmed cell death of lymphoid, myeloid and other haematopoietic, as well as various nonhaematopoietic cell types [1,2]. TNFSF are also implicated in various acquired or genetic diseases, and have been diversely involved in the control of carcinogenesis. While TNFa or TNFR1,2/TNFRSF1A,1B targeting induce tumor development resistance [3,4], TRAIL/TNFSF10 deficiency accelerates hematological malignancies [5]. Furthermore, several TNFSF have demonstrated significant antitumor potential in a number of preclinical murine and human tumor models [6]. Notably, the finding that agonistic 4-1BB monoclonal antibodies can greatly enhance the expansion of CD8 + T cells in vivo and eradicate established tumors raised great interest in 4-1BB as a therapeutic target [7,8]. Hence, strategies that aim to control TNFSF functions are intensively pursued, including the use of recombinant proteins, specific monoclonal antibodies, and by gene transfer. Recently, the tumor-selective action of histone deacetylase inhibitors (HDACi) was shown to involve TRAIL induction in acute myeloid leukemia cells [9,10], paving the road to drug-based antitumor therapeutic strategies targeting TNFSF functions.
HDACi represent a new class of antitumor agents acting on histone deacetylase (HDAC) enzymatic activity [11,12]. Despite the widespread role of HDAC in the transcriptional regulation of gene expression, HDACi are relatively non-toxic to normal cells both in vitro and in vivo, allowing to their use in anti-cancer therapeutic strategies. However, it remains unclear how HDACi exert this cancer cell-selective activity. HDACi are able to induce cell cycle arrest, differentiation, and both intrinsic and extrinsic apoptotic cell death of tumor cells [11][12][13]. Many HDACi have been identified, including the hydroxamate compounds Trichostatin A (TSA) and Suberoylanilide hydroxamic acid (SAHA) that are potent nanomolar HDACi, and the benzamide derivative MS-275, a non hydroxamate micromolar inhibitor of HDACs [13][14][15][16]. Previous reports have shown that these HDACi up-regulate transcription of the p21/WAF/CDKN1A cell-cycle inhibitor and induce TNFSF10 transcription, accounting for the in vitro HDACi-induced G1 arrest and extrinsic apoptosis, respectively [9,10]. In vivo, knocking down TNFSF10 impairs the antitumor effect of MS-275 [10], but antitumor activity of HDACi may further involve additional biological effects such as reduced angiogenesis [17] and inflammation [18], as well as increased immunogenicity [19,20].
In this study, we identified TNFSF9/4-1BBL as a HDACi target that can mediate anti-leukemia allogeneic leukocyte response.

Cells, Culture condition
Jurkat T cells (JA16 clone), Raji cells and U937 cells were maintained in RPMI 1640 medium supplemented with 10% fetal calf serum. Drosophila melanogaster SL2 cells were maintained in Insect-X-Press medium (BioWhittaker) supplemented with 10% FCS at 25uC without CO 2 .

HDACIs and other reagents
Three HDACi were used in this study: Trichostatin A (TSA) was obtained from Sigma, MS-275 and SAHA were obtained from Alexis. Cycloheximide (CHX) and Mithramycin A (MA) were obtained from SIGMA.
Mutations were confirmed by DNA sequencing.

RNA extraction and RT-PCR
Total RNA was isolated from the cells using the RNeasy mini kit (Qiagen) according to the manufacturer's protocol. 2 mg of total RNA was reversed transcribed into cDNA. The expression of TNFSF9 and GAPDH mRNA was analyzed by RT-PCR using the following primers for TNFSF9: forward, 59-GTTTCACTT-GCGCTGCACCTGCAGCCACTG-39 and reverse, 59-TATCA-ACGTCCAACTTGGGGAAGG-39; and for GAPDH, forward, 59-GTCATCCCTGAGCTGAAC-39 and reverse, 59-GGGTCT-TACTCCTTGGAG-39. Amplification was performed with denaturation at 94uC for 50 s, annealing at 60uC (for GAPDH) or 65uC (for TNFSF9) for 45 s and extension at 72uC for 45 s. The PCR products for TNFSF9 (466 bp) and GAPDH (613 bp) were separated by electrophoresis on a 3% agarose gel and visualized by staining with ethidium bromide.

Quantitative RT-PCR
Primer pairs for seventeen TNFSF ligands and twenty-four receptors were incorporated into a low-density array (Assay on Demand, Applied Biosystems). Three endogenous controls were added to the assay set. Three genes were added that represent genes previously described as being either up-or down-regulated by HDACi (see Table 1 for a complete list of genes included in this array). PCR was developed as recommended by the manufacturer. Briefly, 5 ml cDNA (equivalent to 100 ng of total RNA) was mixed with TaqMan Universal Mix (Applied Biosystems) and loaded into 1 sample port. Thermal cycler conditions were as follows: 2 minutes at 50uC, 15 s at 95uC, 60 s at 60uC for 40 cycles. Capture of fluorescence was recorded on the ABI Prism 7900HT scanner, and the C T was calculated for each assay using Sequence Detection System Software 2.1 (Applied Biosystems). Normalization of quantitative-PCR assays was conducted using the C T value of the GAPDH endogenous control. Samples were then converted to a fold change ratio described using standard D C T formula where DC T = C T target -C T average endogenous controls. Thereafter, DDC T values were calculated by subtracting the  [47]. Real time quantitative PCR was performed using a LightCycler rapid thermal cycler (Roche) according to the manufacturer's conditions. Primers for TNFSF9 and GAPDH used were from Applied Biosystems (TaqManH Gene Expression Assays).

Flow cytometry analysis
The cells were treated with HDACi for 4, 8, 24 and 48 h before flow cytometric analysis. Briefly, 2610 5 cells were incubated with anti-mouse IgG-PE (Beckman Coulter) or anti-4-1BBL antibodies (BD Pharmingen) at 4uC for 30 minutes. After three washes with phosphate-buffered saline (PBS) supplemented with 2% FCS, the cells were subjected to flow cytometric analysis. Samples were analyzed on a FACScan (Becton Dickinson) and analyzed by CellQuest software.

Nuclease digestion of purified nuclei and southern blotting
Nuclei were purified and submitted to nuclease digestion as previously described [21]. Briefly, cells were lysed at 4uC in 0.2% NP40 buffer followed by nuclease digestion using DNase I (10 min at 4uC). Proteinase K-treated DNA was next purified by phenol extraction. Purified DNA (30 mg) was digested with Sac I and the fragments generated were separated by electrophoresis. Each size marker was generated by digesting genomic DNA (10 mg) with restriction enzymes as indicated in figure legends. After transfer to nylon membranes (Hybond-N+, Amersham Pharmacia Biotech), DNA was UV-cross-linked and membranes were prehybridized, as previously described [21]. The specific probes were synthesized by PCR using the following primer pairs: 59-CTGCTGATC-GATGGGCCCCTGAGC-39 and 59-GACCTCGGTGAAGG-GAGTCCGGCTG-39. After random primer labeling, denatured DNA probes were allowed to hybridize for at least 16 h at 68uC followed by extensive washes and autoradiography.
Electroporation and Luciferase Assays 10 7 Jurkat T cells were electroporated using Bio-Rad gene pulser II (250 V, 25 ms) with 10 mg of pGL3 or the various TNFSF9 promoter-driven firefly luciferase constructs together with 5 mg of beta-globine Renilla luciferase plasmid (pRL-b). The cells were incubated for 1 h before they were collected, washed in PBS and lysed to determine the luciferase activity by using the dual luciferase reporter assay according to manufacturer's instructions (Promega), and read using a luminometer (Dynex). The transfection efficiency was normalized to Renilla luciferase activity and corrected for protein content as determined by the Bradford protein assay (Bio-Rad). The reported values represent the average of three independent transfections, with standard deviation as errors bars.
SL2 cells (2610 5 ) were plated in 24-well plates and transfection was carried out using Effectene Transfection Reagent (Qiagen). Each well was transfected with 200 ng of reporter plasmid and with pHMW-Sp1 (0 to 600 ng) supplemented, when necessary, with empty vector (pHMW) to equalize the total DNA transfected to 1 mg. Fourty-eight hours after transfection, cells were harvested and assayed for luciferase activity as described above. Human pHMW-Sp1 and its control vector (pHMW) were gifts from Dr Murphy, T.D, Carnegie Institution of Washington, USA.

Allogeneic mixed leukocyte reaction
Raji cells were treated for 24 hours with TSA (250 nM), MS-275 (2.5 mM) or SAHA (2.5 mM). After irradiation, 1610 5 stimulator B lymphomas cells (Raji) were added to 2610 5 PBMCs in each well of 96-well U-bottomed culture plates. OKT3 (0.5 mg/ml) was immobilized onto the surface of the plates (ON, 4uC). Cells were cultured in presence or absence of 1 mg/ml recombinant human 4-1BB-Fc and DR3-Fc (R&D systems) or control Nectine 4-Fc (a gift from M.Lopez). Recombinant proteins were added to the cultures at the same time as cells. Lymphocyte proliferation after 5 days of culture was measured by adding [ 3 H]-thymidine (0,2 mCi/well) to the wells for the last 18 h and the cells were harvested. [ 3 H] incorporation was determined using a TopCount scintillation counter (Packard Instrument).

Cytokine assay
IFN-c levels were measured by ELISA using a cytokine detection kit (BD Biosciences) according to the manufacturer's instructions.

Electrophoretic mobility shift assays
Nuclear extracts were prepared by a rapid method described by Osborn et al [22]. All buffers contained complete protease inhibitors (Roche). Protein concentrations were determined by the method of Bradford [23]. The DNA sequences of the coding strand of the double-stranded oligonucleotides used for this study are listed in Table 2. Electrophoretic mobility shift assays (EMSAs) were performed as described previously [24]. Briefly, nuclear extract (10 mg of protein) was first incubated on ice for 10 min in the absence of probe and specific competitor DNA in a 16 ml reaction mixture containing 10 mg of DNase-free BSA (Amersham Biosciences), 1-2 mg of poly(dI-dC) (Amersham Biosciences) as non-specific competitor DNA, 50 mM ZnCl2, 0.25 mM DTT, 20 mM HEPES (pH 7.3), 60 mM KCl, 1 mM MgCl2, 0.1 mM EDTA and 10% (v/v) glycerol. 20 000 c.p.m. of probe (10-40 fmol) was then added to the mixture with or without a molar excess of an unlabeled specific DNA competitor, and the mixture was incubated for 20 min on ice. Samples were subjected to electrophoresis at room temperature on 6% polyacrylamide gels at 120 V for 2-3 h in 1x TGE buffer (25 mM Tris-acetate, pH 8.3), 190 mM glycine and 1 mM EDTA). Gels were dried and autoradiographed for 24-48 h at 270uC. For supershift assays, polyclonal antibodies against Sp1 (sc-059X), Sp2 (sc-643X), Sp3 (sc-644X), Sp4 (sc-645X), (Santa Cruz Biotechnology), or a purified rabbit immunoglobulin (IgG) were added to the reaction mixture and incubated for 30 min on ice before the addition of the radiolabeled probe.

Up-regulation of TNFSF9 mRNA in HDACi-stimulated leukemia cell lines
Modification of histone and a tubulin acetylation by HDAC inhibition was first investigated in the Jurkat T cell leukemia cell lines using different treatment times and dosages of the hydroxamic-derived Trichostatin A (TSA), the Suberoylanilide hydroxamic acid (SAHA), and the benzamide-derived MS-275 HDACi. As shown in Fig. 1A, while the three HDACi induced dosedependent acetylation of histone H4, which peaked at 16 hours of treatment, only SAHA and TSA induced a tubulin acetylation, as expected from the known incapacity of MS-275 to impact on the specific acetylation of a tubulin by HDAC 6.
These experimental conditions were developped to investigate for the transcriptional modulation of genes of the TNFSF familly using a transcription low-density array (see list of genes in Table 1). Modification of TNFSF mRNA profiles induced by 8 h treatment with TSA, SAHA, and MS-275 were determined in Jurkat cells, followed or not by a 4 h stimulation by a combination of phorbol ester (PMA) and calcium ionophore (Ionomycin) to up-regulate gene transcription. As compared to untreated cells, each HDACi strongly induced p21/WAF/CDKN1A cell cycle inhibitor, used here as control for HDACi-regulated gene transcription (Fig. 1B). Significant (above 2-fold and up to 15-fold) modifications of TNFSF transcript levels were also noted, consistent with previous reports (not shown [12,25]). Interestingly, TNFSF9 was the only TNFSF out of the 17 TNFSF present in the array to be modulated by the three HDACi (not shown). Indeed, as showed in Fig. 1B, the three investigated HDACi up-regulated mRNA levels of TNFSF9 by two-to eight-fold, but not HPRT and b Actin mRNA levels, as normalized to GAPDH levels. Transcriptional induction by the combination of PMA and Ionomycin further increased HDACi-induced TNFSF9 mRNA levels, which was confirmed by RT-PCR analysis (Fig. 1C), as well as by quantitative real time RT-PCR (data not shown). Time course RT-PCR experiments further showed that TNFSF9 mRNA levels were up-regulated by the three HDACi in Jurkat and also in the Raji Burkit B lymphoma and U937 myeloid cell lines, in a time-dependent manner, as compared to endogenous GAPDH levels, yet with distinct kinetics (Fig. 1D).
Up-regulation of TNFSF9 by HDACi does not require de novo protein synthesis and is not associated with DNAse I hypersensitive chromatin remodeling To explore the mechanism of HDACi-induced TNFSF9 upregulation, Jurkat cells were pre-incubated with the protein synthesis inhibitor, cycloheximide (CHX) followed by further incubation for 8 h in the presence or absence of HDACi. As shown in Fig. 2A, TNFSF9 mRNA levels were not reduced by CHX treatment, but instead were observed to be markedly upregulated. This was further confirmed by QRT-PCR (Fig. 2B), suggesting that TNFSF9 mRNA levels are regulated by neosynthesized cellular factors. However, this effect appeared to not be modulated by HDAC inhibition ( Fig. 2A and 2B), indicating that the observed upregulation of TNFSF9 mRNA levels by HDACi does not appear to require neosynthesized cellular factors, and suggests the direct implication of transcriptional regulatory mechanisms. In eukaryotes, chromatin is recognized as an important modulator of transcriptional regulatory mechanisms [26]. Increased histone acetylation, such as following HDAC activity pharmacological inhibition, neutralizes the positive charge of lysine residues, allowing a more ''open'' chromatin structure that facilitates transcription. We thus investigated whether the upregulation of TNFSF9 mRNA levels by HDACiinvolved chromatin remodeling. DNAse I digestion of the TNFSF9 locus followed by indirect end-labeling identified two major dose-dependent DNase I hypersensitive sites (DHSI and II) that were mapped to the 59 regulatory promoter region and between exon 1 and exon 2 (Fig. 2C), respectively. This pattern remained unchanged following HDACi treatment, indicating that the modulation of TNFSF9 mRNA levels by HDACi expression does not involve detectable DNAse I hypersensitive chromatin remodeling within the region under study, however, this study does not exclude localized promoter architecture modifications. The hypersensitivity of the promoter region to DNAse I digestion in unstimulated cells correlated with the basal constitutive transcription of TNFSF9 in Jurkat T cells (data not shown).

Identification and mapping of HDACi responsive elements within the TNFSF9 promoter region
To determine whether HDACi activate TNFSF9 transcription, we isolated human genomic fragments containing the predicted TNFSF9 59 regulatory regions (Fig. 3A). These genomic DNA fragments were inserted upstream to the luciferase reporter gene in the pGL3 basic plasmid construct (Fig. 3B). Promoter activity of the corresponding constructs (pTNFSF9 (1)-(4)) was assayed by measuring firefly luciferase activity after transient transfection in Jurkat T cells (Fig. 3C). pTNFSF9(1), pTNFSF9(2) and pTNFSF9 (3) transcriptional activity was induced up to 12 fold by TSA treatment, indicating the presence of TSA-inducible element(s) within the 985, 398 and 140 bp genomic DNA fragment controlling firefly luciferase transcription (Fig. 3C). In contrast, firefly luciferase activity was not induced by TSA treatment of cells transfected with the pTNFSF9(4) or pGL3 basic empty constructs, indicating that these TSA-inducible element(s) are contained between nucleotide -140 and -65 according to the translation initiation site (Fig. 3). Similar results were observed using different HDACi in different cell lines (data not shown).

Identification of HDACi responsive elements within the TNFSF9 promoter region
Previous studies have shown that conserved cis-regulatory elements, particularly Sp1 and CCAAT boxes, located in the promoters of several genes are responsible for transcriptional activation by HDACi [27][28][29][30][31][32]. Sequence analysis of the pTNFSF9 region contained between the nucleotides 2140 and 265 upstream to the translation initiation site using MatInspector (Genomatix Software, Munich, Germany) and TESS (Transcription Element Search System, CBIL, US) allowed the identification of consensus binding sites for transcriptions factors of the Sp (GC-box) family (Fig. 3A). Mutations of the individual GC-box, as well as combined mutations were performed to evaluate the functional significance of these putative Sp binding sites. As shown in Fig. 4A, individual mutations did not significantly alter basal promoter activity, yet TSA-activated pTNFSF9(3) promoter luciferase activity was reduced by about 25-70% in Jurkat T cells, depending on the individual mutated GC-rich sequence, whereas the combined mutation (pTNFSF9(3) m4) abolished both the basal-and the TSA-induced activity, as compared to the control pGL3 basic reporter construct. Similar results were obtained upon transfection of the pTNFSF9(3) and pTNFSF9(3) m4 promoter constructs in the U937 and Raji cell lines (Fig. 4B). These results identify the GCboxes studied as important mediators of HDACi inducibility of TNFSF9 promoter activity. To gain more insights into the specificity of GC-rich sequences in TSA-induced transcription, a previously described p-486TNFSF6/FasL promoter construct containing    functional GC-rich sequences was transfected into Jurkat cells and analyzed for its responsiveness to TSA treatment ( [21,33]). As shown in Fig. 4C, despite transcriptional induction of this promoter construct by PMA plus Ionomycin treatment, TSA did not stimulate p-486TNFSF6/FasL-mediated transcription. On the opposite, TSA co-treatment down-regulated PMA plus Ionomycin-induced transcription, indicating that the presence of functional GC-rich sequences does not predict TSA-responsiveness (Fig. 4B).

Effect of the mithramycin A DNA intercalating agent on HDACi-induced TNFSF9 gene expression
Mithramycin A is an anti-tumor antibiotic which inhibits transcription from promoters containing GC-rich DNA sequences [34,35]. We thus evaluated HDACi-induced TNFSF9 expression following mithramycin A treatment. As shown in Fig. 4C, mithramycin A prevented TSA-induced TNFSF9 mRNA upregulation. Consistent with the decreased levels of TNFSF9 mRNA, mithramycin A also reduced TSA-induced pTNFSF9(3)mediated transcriptional activity in a dose-dependent manner (Fig. 4D). Collectively, this data strongly indicates that GC-box interacting transcription factors are involved in HDACi-induced TNFSF9 expression.

Sp1/Sp3 transcription factors bind to TNFSF9 promoter GC box elements in vitro and increase its promoter activity
In order to assess the binding of members of the Sp family to the GC-boxes that we identified in the TNFSF9 promoter region, we designed double-stranded oligonucleotides (Fig. 5A). These oligonucleotides were radiolabeled and tested in electrophoretic mobility shift assays (EMSA) for DNA-protein interactions with nuclear extracts from mock-treated and TSA-treated Jurkat cells (Fig. 5B). Retarded protein-DNA complexes were observed using the three distinct GC-box oligonucleotides and TSA treatment did not induce detectable modification of the observed pattern of retarded protein-DNA complexes. To evaluate the sequence specificity of the binding to these GC-boxes, we performed competition EMSAs using different unlabeled double-stranded competitor oligonucleotides (Fig. 5C). The specificity of the protein binding was demonstrated because their formation was inhibited by competition with molar excesses of the unlabeled homologous oligonucleotides (Fig. 5C, lane 3). In contrast, these complexes were not competed by mutated version of the oligonucleotides (lane 4). Similar results were observed for the three GC-boxes. To identify directly the Sp family members within the retarded complexes, we performed supershift assays using specific antibodies directed against individual members of the Sp family of transcription factors (Fig. 5D). The wild-type probe was incubated with nuclear extracts from Jurkat cells and polyclonal antibodies directed against Sp1 and/or Sp3 were added to the binding reaction mixture. The Sp1 antibody selectively supershifted the major slower migrating complex (Fig. 5D, lane 3) and the Sp3 antibody resulted in the strong decrease of the faster migrating complex (Fig. 5D, lane 4). We confirmed these results when both the anti-Sp1 and anti-Sp3 antibodies were included in the same binding reaction (Fig. 5D, lane 5). In contrast, the binding pattern was not affected by the addition of the antibodies directed against other Sp proteins (Sp2 and Sp4) (data not shown), showing that the two complexes did not seem to involve these other proteins. Moreover, the binding pattern was not affected by the addition of purified IgG, used as a negative control (Fig. 5D,  lane 6). Again, similar results were obtained using the three GCboxes. Of note, TSA treatment did not appear to detectably alter Sp1 and/or Sp3 binding (Fig. 5D). Overall, these results demonstrate that Sp1 and Sp3 transcription factors interact with the GC-box (renamed Sp site hereafter in the manuscript) located in the TNFSF9 promoter region. SL2 cells are devoid of endogenous human Sp family transcription factors and thus represent a useful model to investigate Sp-dependent transcriptional mechanisms [36,37]. SL2 cells were cotransfected with the pTNFSF9(3) promoter construct along with pHMW-Sp1 (insect expression vectors encoding for human Sp1). Over expression of Sp1 in SL2 cells resulted in an increased expression of the pTNFSF9(3) promoter construct in a dose-dependent manner ( Figure 5E). Similar results were obtained upon expression of Sp3 transcription factor (not shown).

HDACi-induced TNFSF9 expression increases antileukemia allogeneic leukocyte response
In an attempt to determine the functional significance of HDACiinduced TNFSF9 mRNA levels, 4-1BBL protein expresssion levels and function were evaluated. Time course Flow Cytometry experiments evidenced the up-regulation of 4-1BBL cell surface expression by the three HDACi in Raji cells (Fig. 6A). Ralji cells were next incubated in the presence of HDACi for 24 hours followed by c-irradiation to prevent further cell proliferation and co-culture with purified T lymphocytes from healthy donors. IFNc secretion was determined by ELISA and T lymphocyte proliferation was assessed by [ 3 H]-thymidine uptake. MS-275-stimulated Raji cells significantly up-regulated IFNc secretion by 7 fold (t student, p,0,01), which was prevented by the recombinant 4-1BB-Fc but not control-Fc protein (Fig. 6B), indicating a requirement for 4-1BB/4-1BBL interaction for IFNc secretion stimulated by HDACi-treated Raji cells. TcR/CD3 triggering to activate T lymphocytes to up-regulate cell surface 4-1BB [38][39][40], induced a marked up-regulation of IFNc secretion which was further significantly increased by co-culture of stimulated T cells with MS-275-stimulated Raji cells as compared to untreated Raji cells (6675,86409,3 pg/ml versus 4607,46167,4 pg/ml; t student: p,0,01), in a 4-1BB/4-1BBL interaction dependentmanner (Fig. 6B). In contrast to MS-275, TSA-and SAHAstimulated Raji cells did not stimulate increased IFNc secretion (Fig. 6B). However, HDACi-stimulated Raji cells significantly upregulated T cell proliferation, as compared to untreated Raji cells, using either TSA and SAHA (t student, p,0,05) or MS-275 (t student, p,0,01) (Fig. 6C). Addition of the recombinant 4-1BB-Fc but not control-Fc protein efficiently prevented up-regulation of T lymphocyte proliferation by HDACi-stimulated Raji cells, implicating 4-1BB/4-1BBL interactions in the increased proliferation (Fig. 6C). To further examine for the specificity of 4-1BB/4-1BBL interactions in this HDACi-stimulated mixed lymphoid reaction, DR3-Fc recombinant protein was used as an additional control. Indeed, TL1A/TNFSF15-DR3 interactions have been implicated in T cell proliferation and IFNc secretion [41]. Using the DR3-Fc recombinant protein to block these interactions, only 4-1BB-Fc, but neither the control-Fc nor the DR3-Fc recombinant proteins efficiently prevented up-regulation of T lymphocyte proliferation by HDACi-stimulated Raji cells, specifically implicating 4-1BB/ 4-1BBL interactions in the increased proliferation (Fig. 6E)

Discussion
In this study, we identified 4-1BBL/TNFSF9 as a transcriptionaly activated target of three distinct HDACi in leukemia cell lines and most importantly that MS-275-induced 4-1BB/4-1BBL interactions potentiated anti-leukemia allogeneic leukocyte response, hence providing novel molecular insights into the antileukemia mechanisms of action of HDACi.
Regulation of gene expression is important for HDACi antitumor activities [42][43][44]. Expression profiling studies have identified that limited sets of target genes (2-10%) are regulated by HDACi [45][46][47][48], yet more recent studies have suggested that this number might have been underestimated [43]. Nonetheless, the molecular basis to explain how HDACi impact gene expression remains poorly understood at the genome level. As previously described for CDKN1A/p21/WAF [49], TNFSF10 [10] and other HDACi-activated genes [50], we show here that the TNFSF9 promoter region contains functional Sp1/Sp3 transcrip-  tion factors binding sites that are critical for its HDACi-induced gene transcription. But how Sp1/Sp3 transcription factors modulate TNFSF9 transcription in an HDACi-regulated manner remains to be established. Using the indirect-end labeling technique, DNAse I hypersensitive sites (DHSI and II) were identified, DHSI being mapped to the TNFSF9 minimal promoter region containing the Sp1/Sp3 DNA binding sites. DHSI was detected in unstimulated Jurkat cells displaying basal TNFSF9 transcription and was not detectably altered upon HDACi treatment, in agreement with the transcriptional activation of TNFSF9 promoter constructs by HDACi and suggesting that HDACi-induced TNFSF9 transcription did not involve DNase I-sensitive large chromatin remodeling events. The transcriptional activity of Sp1 family members can be modulated by acetylation [13,51,52] and both Sp1 and Sp3 can interact with HDAC [53], suggesting a role for Sp1 and/or Sp3 acetylation in the Sp1/Sp3-dependent, HDACi-regulated TNFSF9 induction. Alternatively, Sp1/Sp3 may recruit acetylated co-factors allowing TNFSF9 transcriptional induction. To asses the in vivo binding of Sp factors to the TNFSF9 promoter region, we attempted to perform chromatin immunoprecipitation assays. Preliminary experiments showed low enrichement of the TNFSF9 promoter region in Sp1 immunoprecipitates, as compared to control antibodies, that were also not detectably modified by TSA treatment in agreement with previous observations (( Nebbioso, 2005 #173] and data not shown).
Of interest, nonetheless, mithramycin A abrogated both HDACi-stimulated endogenous and promoter-driven TNFSF9 transcription, yet it did not affect basal TNFSF9 transcription, supporting the hypothesis that acetylation of Sp1/Sp3 or associated factors, and not per se Sp1/Sp3 chromatin-recruitment, was required for HDACi-regulated TNFSF9 induction. Consistent with this hypothesis, EMSA experiments did not reveal increased Sp1/Sp3 DNA binding upon HDACi treatment. Collectively, these results support a model where protein acetylation induced by HDACi increases the transcriptional activity of Sp1/Sp3-dependent complexes seated on the TNFSF9 promoter. The mechanisms of acetylation of Sp1/Sp3 or associated factors in HDACi-induced TNFSF9 transcription remains to established.
One important finding of the present study was the functional up-regulation of 4-1BBL by MS-275, as evidenced by enhanced T cell proliferation and IFNc production in mixed peripheral blood leukocytes co-culture experiments. While TRAIL upregulation was convincingly shown to contribute to the selective HDACi-induced apoptosis of tumor cells, our results provide first evidence that up-regulated 4-1BBL might further contribute to the in vivo anticancer action of HDACi such as MS-275. There has been ample evidence demonstrating the robust in vivo antitumor immune responses upon 4-1BB triggering [6,7]. This effect is largely interpreted by 4-1BB signaling on tumor-specific T cells that can enhance proliferation and CTL activity, and can prevent activation-induced cell death [8,40]. Consistently, co-cultures of HDACi-treated Raji cells with T cell receptor-stimulated peripheral blood leukocytes enhanced T cell proliferation as compared to untreated Raji cells. This proliferative response was blocked in the presence of competing 4-1BB-Fc, but not by DR3-Fc recombinant protein, showing that 4-1BB/4-1BBL interaction was specifically required for this HDACi-enhanced T cell proliferative response. More recent studies have further revealed additional mechanisms that may account for 4-1BB anti-tumor effects. Firstly, 4-1BB signaling is able to prevent and rescue T cells from immune tolerance, promoting regression of poorly immunogenic tumors [54]. Secondly, 4-1BB expression on non-T cell populations, including natural killer cells and dendritic cells has been implicated [55,56]. Finally, 4-1BB stimulation modifies the distribution pattern of tumor-specific T cells in vivo, in a manner that is largely dependent on IFNc [57]. IFNc, a pleiotropic cytokine secreted by activated T cells and NK cells, plays a central role in both the innate and adaptive immune response to a variety of pathogens and transformed cells [58]. CD4 + and CD8 + T cells have been shown to mediate tumor rejection in an IFNc-dependent fashion through different mechanisms, including up-regulation of molecules critical for antigen processing and presentation [59], differentiation of fully competent effector cells [60], secretion of angiostatic chemokines by stromal cells present within a tumor [61], and regulation of Tcell migration to the tumor site [62]. In this study, co-cultures of MS-275-treated Raji cells with T cell receptor-stimulated peripheral blood leukocytes enhanced T lymphocyte IFNc secretion, as compared to untreated Raji cells. Increased IFNc secretion was blocked in the presence of competing 4-1BB/ TNFRSF9-Fc recombinant protein, showing that 4-1BB/4-1BBL interactions were required. Importantly, induction of IFNc was also observed in the absence of T cell activation, and was completely blocked in the presence of the 4-1BB-Fc recombinant protein, indicating that induction of IFNc by MS-275-treated Raji cells occurred through 4-1BB triggering and could occur in the absence of T cell activation.
Despite similar levels of induction of 4-1BBL transcription and cell surface expression by TSA, SAHA and MS-275, only MS-275 potentiated IFNc secretion in mixed leukocyte reactions and also provided more efficient T lymphocyte proliferative responses in these assays. These results suggest that compared to TSA and SAHA, MS-275 can provide additional specific signalling to improve further immune responses. Previous reports have shown that HDACi can induce the expression of CD86/B7.2 costimulatory molecule in acute myeloid leukemia (AML) cells and freshly isolated AML clinical samples [19]. Taken together with our present finding that HDACi can enhance expression of yet another costimulatory molecule, 4-1BBL, this data further supports HDACi, especially MS-275, as antitumor drugs to be used in immunotherapeutic clinical approaches. HDACi have been shown to induce growth arrest, differentiation, and/or apoptosis of cancer cells in vitro and in vivo tumor-bearing animals models [13,63,64]. In association with the proapoptotic effect of HDACi, the enhancement of costimulatory molecules on leukemia blasts could augment tumor immunogenicity, increasing specific CTL activity against tumor cells. Combinatorial treatment modulating both apoptotic and costimulatory molecules may be more effective for reinforcing host immunity and eradicating tumors. This point is supported by a recent study by Uno et al. who showed that induction of tumor cell apoptosis by an agonistic antibody to TRAILR2/DR5, combined with T cell activation by CD40 and 4-1BB agonistic monoclonal antibodies, potently and rapidly stimulated tumor-specific CD8 + T cells capable of eradicating pre-established tumors [65]. It is noteworthy that HDACi treatment did not induce 4-1BBL transcription in breastderived tumor cells lines, nor did it induce significant 4-1BBL transcription in primary human peripheral blood lymphocytes, suggesting leukemia specific modulation of 4-1BBL by HDACi treatment (BV, data not shown).
In conclusion, we have demonstrated that HDACi can modulate the expression of 4-1BBL costimulatory molecule in B leukemia cell line model, which can enhance T cell responses towards the tumor cells. Further studies are now required to examine these effects using primary cells from patients with chronic lymphocytic leukemia (CLL) and to determine in animal models whether 4-1BBL/TNFSF9 up-regulation contributes to HDACi-induced anti-tumor effects in vivo. The distinct capacity of MS-275 as compared to other HDACi to modulate these T cell responses will also require further investigation.