Tax Protein-induced Expression of Antiapoptotic Bfl-1 Protein Contributes to Survival of Human T-cell Leukemia Virus Type 1 (HTLV-1)-infected T-cells

Tax Protein-induced Expression of Antiapoptotic Bfl-1 Protein Contributes to Survival of Human T-cell Leukemia Virus Type 1 (HTLV-1)-infected T-cells. Background: Antiapoptotic Bcl-2 proteins are up-regulated in HTLV-1-infected T-cells. Results: Antiapoptotic Bfl-1 protein is regulated by viral Tax and HBZ proteins and is involved in survival of HTLV-1-trans-formed T-cells. results directly implicate and in HTLV-1-in-fected T-cell survival and suggest that both Bfl-1 and Bcl-x L represent potential therapeutic targets for ATLL treatment.

Adult T-cell leukemia/lymphoma (ATLL) 6 is associated with human T-cell leukemia virus type 1 (HTLV-1) infection. It is an aggressive malignancy of mature activated CD4 ϩ T-cells and in rare cases of CD4 ϩ CD8 ϩ or CD8 ϩ T-cells. Among the 10 -20 million people infected with HTLV-1 worldwide, 2-10% will develop an ATLL after a long period of clinical latency. ATLL is characterized by the persistence and clonal expansion of CD4 ϩ -infected T-cells (1,2), but the molecular mechanisms of ATLL development are complexes and remain incompletely understood. Some of these mechanisms are related to the expression of the regulatory viral Tax protein and of the HTLV-1 basic leucine zipper (bZIP) factor (HBZ) (3). The oncoprotein Tax is highly expressed in early stages of ATLL development, although its expression decreases to become almost undetectable in most ATLL cells (4 -6). Strong evidence indicates that Tax plays a central role in T-cell transformation both in vitro in human cells and in vivo in transgenic animal models. However, Tax-induced immortalization of human primary T-cells is a very rare event (7)(8)(9)(10)(11)(12)(13). Tax modulates cellular gene expression and interferes with the control of cell survival, proliferation, and genetic stability of infected cells (14 -16). Tax does not directly bind to DNA, but it promotes the recruitment of transcription factors on targeted cellular genes. In particular, Tax activates survival transcription factors, such as nuclear factor-B (NF-B) and activator protein-1 (AP-1) members, which in turn increase the expression of antiapoptotic proteins (16 -21). As a consequence, both HTLV-1-and Tax-transformed T-cells show higher resistance to apoptosis than untransformed control cells (18,22).
Although Tax is not detectable in 60% of ATLL cases, HBZ remains expressed through all stages of the ATLL process (14,15,(23)(24)(25)(26)(27). HBZ controls gene transcription by interacting with Jun members of the AP-1 family through their bZIP domain (28). An increasing number of studies report that HBZ promotes T-cell proliferation and inflammation and suggest that HBZ participates in the maintenance of tumoral phenotype (23, 28 -32).
Many viruses responsible for the development of leukemia/ lymphoma have evolved to escape immune surveillance. Some of them inhibit apoptosis by encoding viral Bcl-2 (B-cell lymphoma gene-2) analogs, which mimic their cellular antiapoptotic function. Others up-regulate the expression of cellular antiapoptotic Bcl-2 proteins (33)(34)(35). Impaired apoptosis associated with an imbalance of the expression of Bcl-2 members in favor of antiapoptotic proteins is a hallmark of human hematopoietic malignancies and is frequently associated with resistance to therapy (36,37). The NF-B transcription factors have been shown to directly up-regulate both bfl-1 (Bcl-2 fetal liver) and bcl-x L (Bcl-2-like long) gene expression, and overexpression of both proteins has been associated with increased resistance of tumor cells to apoptotic stimuli or to chemotherapeutic drugs (38 -47). Interestingly, several data point out a potential role for NF-B-mediated bfl-1 and bcl-x L overexpression, by viral proteins, in both B-and T-lymphoma/leukemia. Indeed, the bfl-1 promoter is activated by EBV latent membrane protein 1 (LMP1), as well as by EBV nuclear antigen 2 (EBNA2) and HTLV-1 Tax proteins (48 -50). Similarly, bcl-x L gene is activated by EBV LMP2A and HTLV-1 Tax proteins in B-and T-cells, respectively (51)(52)(53). Although the regulation of bcl-x L by Tax protein has already been documented, the mechanism underlying the regulation of bfl-1 expression by viral Tax and HBZ proteins and the involvement of Bfl-1 in HTLV-1infected T-cell survival remain unknown.
Here, we report that Bfl-1 is expressed in HTLV-1-infected T-cell lines but not in uninfected T-cells. We demonstrated that Tax induces Bfl-1 expression through the canonical NF-B pathway but also synergizes with JunD or c-Jun of the AP-1 family to activate bfl-1 transcription. By contrast, HBZ modulates Jun-mediated bfl-1 gene activation. Moreover, both NF-B and AP-1 bind to different sites of the bfl-1 promoter in a T-cell line stably expressing tax, hbz, and bfl-1 transcripts. Finally, we showed that knockdown of Bfl-1 or Bcl-x L , but not Bcl-2, decreases HTLV-1-infected T-cell survival and that targeting both Bfl-1 and Bcl-x L restored full cell death. Altogether, our data strongly suggest that Bfl-1 and Bcl-x L represent potential therapeutic targets for ATLL treatment.
Cells-HeLa cell line was grown in DMEM (Invitrogen) supplemented with 10% endotoxin-free FCS (Lonza), 2 mM L-glutamine, 10 mM HEPES, and 40 g/ml gentamicin (Invitrogen). The HTLV-1-infected MT-2, MT-4, C8166-45, SP, and FC36.22 T-cell lines were obtained from AIDS Research and Reference Reagent Program. The HTLV-1-infected T-cell lines C91PL and HuT-102 were provided by E. Wattel (Lyon, France) and R. Mahieux, respectively. Uninfected human leukemia Jurkat (Clone E6 -1), Sup-T1, Molt-4 (Clone 8) and CEM, and lymphoma HuT-78 T-cell lines, which served as negative controls, were obtained from AIDS Research and Reference Reagent Program. JPX-9 cells are derivatives of Jurkat, which have a stably integrated tax gene under the control of a metallothionein promoter and are kindly provided by M. Nakamura (Tohoku University, Japan (62)). To induce Tax expression, JPX-9 cells were cultured in the presence of 20 M CdCl 2 . Stably Tax/HBZ-expressing Jurkat (E12 clone) T-cells were obtained as reported previously (63). The T-cell lines were grown in RPMI 1640 medium (Invitrogen) supplemented as for DMEM medium. FC36.22 and SP cells were cultured in the presence of 4% of homemade IL-2.
We used frozen peripheral blood mononuclear cells (PBMC) from healthy donors (Etablissement de Transfusion Sanguine, Lyon, France) and ATLL patients (Prof. O. Hermine, H. Necker, Paris, France). The clinical status of the ATLL patients at the time of sample collection is summarized in supplemental Table  S1. Fresh human PBMC were stimulated with mAbs anti-CD3/ CD28 (Dynabeads Human T-cell activator, Invitrogen) 48 h prior to transduction. Then the Tax-expressing T-cells were obtained by transduction of stimulated PBMC with a lentivirus vector encoding a HTLV-1 Tax-enhanced YFP fusion, as described previously (64). The obtained CD4 ϩ Tax-T-cells were grown in complete RPMI medium supplemented with 1% (v/v) Eagle's minimum essential medium nonessential amino acid, 1 mM sodium pyruvate, and 200 units/ml human IL-2 (PeproTech). Untransduced T-cells from blood were sorted by FACS according to CD4 and CD8 expression using FACSVantage SE option Diva (BD Biosciences). Purity was over 90%.
RNA Extraction and Real Time RT-PCR-0.5-2 ϫ 10 6 T-cells were lysed using TRIzol reagent (Invitrogen), and cDNAs were obtained as described previously (65). 0.1-0.5 ϫ 10 6 transfected HeLa cells were lysed according to the manufacturer's specifications (SV total RNA isolation kit, Promega). The real time RT-PCR was performed with FastStart SYBR Green Master (Roche Applied Science) on a StepOne Plus machine (Applied Biosystems) using 4 M of each primer (supplemental Table S2). cDNA levels were normalized using the housekeeping HPRT and GusB genes. Each sample was run in triplicate, and the data were analyzed as described previously (65).
Immunoblotting Analysis-2-5 ϫ 10 6 cells were lysed in 50 l of Nonidet P-40 lysis buffer (150 mM NaCl, 20 mM Tris-HCl (pH 7.4), 2 mM EDTA, and 1% Nonidet P-40) supplemented with protease inhibitor mixture (Roche Applied Science). Cytoplasmic protein extracts (70 -100 g) were separated on a NuPAGE 4 -12% BisTris gel (Invitrogen) and transferred to nitrocellulose membrane (Invitrogen). Protein expression was analyzed by immunoblotting using specific Abs. Protein-Ab complexes were visualized by chemiluminescence (Western Lighting Chemiluminescence Reagent Plus; PerkinElmer Life Sciences). Transduction Assays-RNA interference sequences are summarized in supplemental Table S3. Corresponding oligonucleotides were cloned into the lentiviral vector pLL3.7 co-expressing GFP as described previously (47). 3 ϫ 10 5 cells were cultured in a 6-well plate for 24 h. The cells were then transduced with lentiviral vector expressing specific shRNA at 16 multiplicities of infection in the presence of Polybrene (Sigma), as performed previously (47). Cells were collected at days 5-7 to evaluate efficiency of transduction (GFP ϩ ) by flow cytometry, mRNA, or protein extinction and mortality by flow cytometry after staining with 0.5 g/ml propidium iodide (PI ϩ ).
Transfections and Luciferase Assays-2.5 ϫ 10 5 or 1.25 ϫ 10 5 HeLa cells were plated on 6-or 24-well plates overnight, respectively. Then, cells were transiently transfected with 3 g/6-well or 1 g/24-well plasmids using JetPEI according to the manufacturer's protocol (Ozyme). 48 h post-transfection, cells were collected for RNA and protein extractions or for luciferase assays. To compare the amounts of expressing vectors between experiments performed in 6 and 24 wells, amounts are presented in the figures as nanograms or micrograms of interest plasmid/g of total transfected vectors. Luciferase activities were determined in triplicate using a luciferase assay kit system (Dual-Glo, Promega) and a luminometer (Infinite M200, Tecan) according to the manufacturer's specifications. Lucifer-ase activity was calculated using the ratio Luc/Ruc. Normalized luminescence values are presented as relative fold trans-activation by comparison with the basal bfl-1 luciferase activity obtained using empty vector, which was arbitrarily set up to one.
Electrophoretic Mobility Shift Assay (EMSA) and Supershift-For the preparation of nuclear extracts, 10 7 cells were washed with PBS, and the volume of the pellet was determined. Cells were resuspended in a 4-pellet volume of cold lysis buffer (20 mM Tris-HCl (pH 8), 1 mM EDTA, 5 mM DTT) with protease inhibitors (Sigma). After 15 min on ice, cells were lysed by homogenization in a loose-pestle Dounce, and the mixture was centrifuged at 5,000 ϫ g at 4°C for 10 min in an Eppendorf microcentrifuge. The pellet was resuspended in 4 pellet volumes of nuclear lysis buffer (20 mM Tris-HCl (pH 8), 20% glycerol, 0.42 M NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM DTT) with protease inhibitors and incubated on ice for 15 min. The mixture was then homogenized in a tight pestle Dounce and centrifuged 10,000 ϫ g at 4°C for 10 min. The resulting supernatant was used as the nuclear extracts in EMSA.
Cell Death Assay-T-cell lines (4 ϫ 10 5 /ml) were cultured in the absence or presence of various concentrations of DMSO, ABT-737 (Euromedex, Selleck Chem) (66) or etoposide (Sigma). 48 h after treatment, mortality was determined by flow cytometry after staining with PI or with allophycocyanin-con-jugated rabbit anti-active caspase-3 antibody after fixation and permeabilization using the BD Cytofix/Cytoperm TM kit.
Statistical Analysis-Statistical analysis of the differences between the groups was performed using t test with p Ͻ 0.05 considered statistically significant.  Table S4). Interestingly, Bfl-1 mRNA and proteins were detected in all HTLV-1-infected T-cell lines regardless of their phenotype but not in uninfected T-cell lines, except very moderate mRNA levels in HuT-78 ( Fig. 1, A and B). Note that Bfl-1 expression was observed in IL-2-dependent (SP and FC36.22) and -independent (MT-4, MT-2, HuT-102, and C91PL) HTLV-1-infected T-cells (Fig. 1, A and B). We also compared the expression of Bcl-x L and Bcl-2 in these T-cell lines. In agreement with previous reports (52,53), we observed that Bcl-2 and Bcl-x L messengers and proteins were expressed at different levels in both uninfected and HTLV-1-infected T-cell lines (supplemental Fig. S1, A and B). The presence of HTLV-1 was confirmed by the expression of Tax and HBZ (Fig.  1, A and B, and supplemental Fig. S1A). Altogether, these data show that, in contrast to Bcl-2 and Bcl-x L , Bfl-1 expression is restricted to HTLV-1-infected CD4 ϩ CD8 ϩ , CD8 ϩ , and CD4 ϩ T-cells.

Antiapoptotic Bfl-1 Protein Is Overexpressed in HTLV-1-infected T-cells-De
We then examined the expression of a Bfl-1 protein in PBMC from three healthy donors and four ATLL samples. Although healthy PBMC expressed between 25 and 46% of CD3 ϩ CD4 ϩ T-cells, all PBMC from ATLL patients exhibited more than 70% of CD3 ϩ CD4 ϩ T-cells (supplemental Table S5). As expected, only the CD4 ϩ ATLL cells showed an activated phenotype. Interestingly, Bfl-1 protein was detected in two acute aggressive ATLL cells but not in two chronic indolent ATLL cells and in healthy control cells ( Fig. 1C and supplemental Table S1).
Tax Activates bfl-1 Gene Transcription in T-cells-To determine the molecular mechanism of Bfl-1 induction by HTLV-1, we first focused on Tax protein, which is known to be involved in T-cell survival. We used JPX-9 cells, a Jurkat subclone that carries the tax gene under the control of the inducible metallothionein promoter. Treatment of JPX-9 cells with CdCl 2 induced tax mRNA levels within 12 h, and tax expression persisted until 48 h ( Fig. 2A). We observed that bfl-1 transcript levels were significantly induced only 24 h post-treatment with CdCl 2 and further increased with time. These results indicated that tax expression preceded that of bfl-1 and that the presence of bfl-1 correlated with that of tax. Then we measured bfl-1 mRNA expression in human primary CD4 ϩ T-cells. We observed that CD4 ϩ T-cells expressing tax mRNA after transduction expressed higher levels of bfl-1 mRNA, compared with untransduced T-cells (Fig. 2B). Note that untransduced T-cells and Tax-transduced CD4 ϩ T-cells expressed similar levels of bfl-1 mRNA to uninfected HuT-78 and HTLV-1-infected MT-2 and C91PL CD4 ϩ T-cells, respectively (Figs. 1A and 2B).
To further demonstrate that Tax activates bfl-1 gene expression, we knocked down the expression of Tax using the lentivirus-based vector pLL3.7 expressing Tax-specific shRNA in stably Tax-expressing E12 T-cell clone. Tax shRNA induced more than 80 and 60% of inhibition of tax and bfl-1 mRNA expression, respectively, compared with Ctl shRNA (Fig. 2C). Thus, these data indicate that Tax induces bfl-1 gene expression in T-cells. Tax trans-Activates the Transcription of bfl-1 Gene via the Canonical NF-B Pathway-As Tax trans-activates the expression of numerous genes via the NF-B pathway (20), and as bfl-1 promoter contains NF-B-binding sites (38, 50, 67), we then investigated the role of the NF-B pathway in Tax-medi-ated bfl-1 transcription (Fig. 3). HeLa cells were transfected with expressing vectors encoding for either WT Tax (Tax-WT), Tax-M47 (⌬CREB/ATF), or Tax-M22 (⌬NF-B) mutants (Fig.  3A). The tax and bfl-1 mRNA levels were then measured by real time RT-PCR (Fig. 3B). Although mRNA levels of Tax-WT and Tax mutants were quite similar (Fig. 3B, left panel), only Tax-WT and Tax-M47 mutant efficiently induced bfl-1 mRNA expression (Fig. 3B, right panel). By contrast, Tax-M22 mutant induced only a weak bfl-1 mRNA expression, suggesting that NF-B pathway is involved.
To further confirm the involvement of the NF-B pathway in Tax-mediated bfl-1 transcription, we used the dominant-negative IB␣ S32A/S36A (IB␣DN) mutant construct that prevents the phosphorylation and thus the degradation of IB␣, an inhibitor of the canonical pathway. HeLa cells were co-transfected with empty-or Tax-expressing vectors, bfl-1-luciferase promoter plasmid (Bfl-1-Luc), and increasing amounts of IB␣DN-expressing vectors. As shown in Fig. 3C, overexpression of IB␣DN resulted in a dose-dependent reduction of the Tax-mediated bfl-1 promoter activation (Fig. 3C, upper panel). Expression of IB␣ and Tax proteins was confirmed by immunoblotting (Fig. 3C, lower panel). We observed that IB␣DN was unable to totally inhibit the Tax-mediated bfl-1 promoter activation even when over 100 ng of IB␣DN-expressing vector were transfected (Fig. 3C, upper panel, and data not shown). By contrast, IB␣DN completely inhibited Tax-mediated NF-Bresponsive promoter when used at 25 ng (supplemental Fig. S2, left panel). As expected, IB␣DN had no effect on HTLV-1-LTR trans-activation by Tax (supplemental Fig. S2, right panel), showing that this inhibitor was specific to the NF-B pathway. Finally, we analyzed the effect of IB␣DN on bfl-1 messenger expression mediated by Tax. HeLa cells were co-transfected with Tax-WT-and IB␣DN-expressing vectors, and bfl-1 transcript levels were measured by real time RT-PCR. Overexpression of IB␣DN resulted in a 70% inhibition of the Tax-mediated endogenous bfl-1 mRNA expression (Fig. 3D).
Altogether, these findings demonstrate that Tax trans-activates the bfl-1 gene through the NF-B pathway, but it also suggests that other transcription factors might be involved in this regulation. Reinforcing such an hypothesis, we observed that the NF-B inhibitor Bay11-7082 reduced bfl-1 mRNA level in HTLV-1-infected C91PL and MT-4 T-cell lines (data not shown).
c-Jun and JunD Cooperate with Tax to Activate bfl-1 Gene Transcription-Previous reports showed that the Tax-M22 mutant, deleted for its NF-B binding ability, was also unable to activate the AP-1-responsive promoter (68). Furthermore, the bfl-1 promoter contains AP-1-binding sites (50,67). Therefore, we tested whether AP-1 may also be involved in trans-activation of the bfl-1 promoter by Tax. We first investigated the role of AP-1 transcription factors on the activation of bfl-1 gene expression by performing luciferase reporter assays. Co-transfection of HeLa cells with either JunB, c-Jun, or JunD, and bfl-1-luciferase promoter-expressing vectors showed that both c-Jun and JunD activated the bfl-1 promoter in a dose-dependent manner, although JunB did not (supplemental Fig. S3). We next analyzed the potential cooperation between either Tax and c-Jun or Tax and JunD using suboptimal amounts of AP-1-

Tax of HTLV-1 Mediates T-cell Survival via Bfl-1 and Bcl-x L
expressing vectors. Interestingly, overexpression of Tax with c-Jun or JunD resulted in an increased activity of the bfl-1 promoter activity in a dose-dependent manner, compared with Tax, c-Jun, or JunD alone (Fig. 4, A and B, upper panels). Immunoblotting experiments confirmed the expression of Tax, c-Jun, or JunD proteins in transfected HeLa cells (Fig. 4, A and B, lower  panels). To corroborate these results in T-cells, the Jurkat T-cell line was nucleofected with the bfl-1 promoter expressing vector and empty-, Tax-, and c-Jun-or JunD-expressing vec-tors. As expected, we observed that Tax cooperated with c-Jun or JunD in Jurkat T-cells to activate the bfl-1 promoter (data not shown). We next examined the cooperation of Tax with either c-Jun or JunD on endogenous bfl-1 gene expression. For that purpose, HeLa cells were co-transfected either with Tax-and c-Jun-expressing or Tax-and JunD-expressing vectors, and bfl-1 mRNA expression was evaluated by real time RT-PCR. We observed that bfl-1 gene transcription was highly increased in cells co-expressing Tax and c-Jun or Tax and JunD as com- pared with cells expressing only Tax, c-Jun, or JunD (Fig. 4C). Thus, these results show for the first time that Tax acts synergistically either with c-Jun or JunD to trans-activate the bfl-1 gene. (29,57,58,69), we assessed on the ability of HBZ viral protein to regulate bfl-1 gene activation. We first analyzed the role of HBZ on the activation of bfl-1 gene expression by performing luciferase reporter assays. Co-transfection of HeLa cells with HBZ-expressing vector (Fig. 5A) and bfl-1-luciferase promoter plasmid showed that HBZ was unable to activate the bfl-1 promoter, whatever the amount of HBZ expressing vector used (Fig. 5, B  and C, upper panels). We next analyzed the potential effect of HBZ on c-Jun-or JunD-mediated bfl-1 promoter activation using optimal amounts (25 ng) of Jun-expressing vectors. Interestingly, overexpression of HBZ inhibited c-Jun-mediated activation of the bfl-1 promoter in a dose-dependent manner (Fig.  5B, upper panel). By contrast, overexpression of HBZ with JunD resulted in the amplification of JunD-mediated bfl-1 promoter  activation in a dose-dependent manner (Fig. 5C, upper panel). Immunoblotting experiments confirmed the expression of HBZ, c-Jun, or JunD in transfected HeLa cells (Fig. 5, B and C,  lower panels).

HBZ Regulates Jun-induced bfl-1 Promoter Activation via Both Activation and Basic Leucine Zipper Domains-As HBZ modulates gene expression via c-Jun and JunD
We further examined the role of activation domains and bZIP domains in HBZ modulating Jun-induced bfl-1 promoter activation, using the well characterized HBZ mutants (HBZ-⌬AD and -⌬bZIP) (Fig. 5A). Vectors expressing HBZ-WT or its deleted forms and the bfl-1 promoter were transiently co-transfected into HeLa cells with or without c-Jun-or JunD-expressing vectors. As shown in Fig. 5D, HBZ⌬AD partially repressed c-Jun-mediated bfl-1 promoter activation, whereas HBZ⌬bZIP was unable to do so (left panel). By contrast, HBZ⌬AD was unable to enhance JunD-mediated bfl-1 promoter activation, whereas HBZ⌬bZIP had no significant effect on JunD-mediated bfl-1 promoter trans-activation (Fig. 5D, right panel).
Altogether these results demonstrate the following: (i) HBZ suppresses c-Jun-mediated bfl-1 promoter activation; (ii) HBZ acts synergistically with JunD to increase the activation of the bfl-1 promoter; and (iii) both the activation and basic leucine zipper domains of HBZ are required to modulate c-Jun-or JunD-induced bfl-1 promoter activation.

NF-B, c-Jun, and JunD Binding Activity to the bfl-1 Promoter in T-cells-We have shown that Tax induced bfl-1 gene activation via both NF-B and AP-1 pathways and that HBZ modu-
lates Jun-mediated bfl-1 promoter activity. To determine the resulting activity of both Tax and HBZ on the endogenous expression of bfl-1 transcripts, we measured bfl-1 mRNA levels by real time RT-PCR in the E12 cells, a clone derived from Jurkat T-cells that stably express Tax and HBZ, and in C8166-45 T-cells carrying but not expressing the HTLV-1 viral proteins, except Tax and HBZ. As shown in Fig. 6A, both E12 and C8166-45 T-cells expressed a high level of tax and a low level of hbz transcripts. More interestingly, bfl-1 is highly expressed in these T-cells (Fig. 6A), suggesting that Tax may have a dominant-positive effect when its expression is higher than that of HBZ.
We next determined whether Tax and/or HBZ are responsible for NF-B and/or AP-1 binding to respective consensus elements in the proximal part of the bfl-1 promoter. EMSA was performed with double-stranded oligonucleotides representing the NF-B at position Ϫ52 and AP-1 elements at positions Ϫ104 or Ϫ74. Consistent with the luciferase assays, protein complexes bound to both NF-B (Fig. 6B, lane 2) and AP-1 (Fig.  6B, lanes 6 and 10 for AP-104 and AP74, respectively) sites were detected with nuclear extracts from E12 T-cell clone. The specificity of DNA-protein complexes in extracts was determined by competition studies with unlabeled competitors. As expected, cold B52, AP-104, and AP74 double-stranded oligonucleotides efficiently competed with their homologously labeled probes and decreased binding of nuclear extracts in a dose-dependent manner (Fig. 6B, lanes 3 and 4 for B52, lanes  7 and 8 for AP-104, and lanes 11 and 12 for AP74). Then, the exact composition of the transcription factor DNA-protein complexes in E12 T-cell clone was ascertained using antibody supershifting EMSA. Supershift analyses were performed on agarose gels to achieve a better resolution of the antibody-protein-DNA complexes. Antibodies against p65/RelA, c-Rel, RelB, p100/p52, and p105/p50 all induced a supershift of the DNA-protein complexes using the B52 site of the bfl-1 gene as probe (Fig. 6C, lanes 2-6, respectively) with a variable degree of shift depending on the antibody used. Antibodies against c-Jun and JunD induced a supershift of complexes formed with either an AP-104 or AP-74 DNA-binding site (Fig. 6C, lanes 8 and 9  and 11 and 12, respectively). Note that isotype controls were unable to induce supershift, demonstrating the specificity of interactions between NF-B and AP-1 with the bfl-1 promoter (data not shown). Taken together, our data indicate that NF-B, c-Jun, and JunD proteins bind to the proximal part of the B and AP-1 elements of the bfl-1 promoter in E12 T-cells expressing Tax, HBZ, and Bfl-1. Moreover, these results suggest that both canonical and alternative NF-B pathways are involved in the B binding activity of the bfl-1 promoter.
Bfl-1 Is Involved in HTLV-1-infected T-cell Survival-We finally asked whether Bfl-1 participates in the survival of HTLV-1-infected T-cell lines. As the impact of the antiapoptotic Bcl-2 proteins in HTLV-1-infected T-cell survival has never been studied, we also compared the involvement of Bfl-1 to that of the antiapoptotic Bcl-2 or Bcl-x L proteins. We thus knocked down the expression of these proteins using a lentivirus-based vector pLL3.7 that co-expressed specific shRNA and GFP as reporter genes, thus permitting transduced cells to be tracked by flow cytometry. After 5-7 days of infection, the transduction efficiency of uninfected (Jurkat) and HTLV-1-infected (C91PL and MT-4) T-cell lines ranged from 87 to 98% (data not shown). Bcl-2, Bcl-x L , and Bfl-1 silencing by the specific shRNAs was confirmed by immunoblotting (Fig. 7A). We observed that Bcl-x L and Bfl-1 knockdown induced a moderate death of C91PL T-cells (Fig. 7B, middle panel) with only 20 and 40% of cell death detected in C91PL cells transduced with Bcl-x L and Bfl-1 shRNA, respectively, suggesting that Bfl-1 expression may offset the Bcl-x L knockdown and conversely. In contrast, we observed that MT4 T-cells were highly sensitive to Bfl-1 silencing, whereas cell death resulting from Bcl-x L knockdown never exceeded 25% after 7 days in these cells (Fig. 7B, right panel). Finally, Ctl or Bcl-2 shRNA did not induce significant death in transduced C91PL and MT-4 T-cells, and the viability of uninfected Jurkat T-cells was not altered by any shRNA tested (Fig. 7B). Altogether, these data directly involve Bfl-1, but also Bcl-x L , in the survival of HTLV-1-transformed T-cell lines.

HTLV-1-infected T-cell Survival Is Fully Compromised by Combination of Bfl-1 Knockdown and ABT-737 Treatment-As
Bcl-x L and Bfl-1 knockdown induced, respectively, only 20 and 40% of cell death in C91PL (Fig. 7B), we assessed whether targeting both Bfl-1 and Bcl-x L proteins might compromise C91PL cell survival. Among the small compounds that target Bcl-2 antiapoptotic proteins, ABT-737 has been described to specifically block Bcl-2, Bcl-x L , and Bcl-w activity, but not Bfl-1 and Mcl-1 activity (66). We first explored survival of the C91PL T-cell line after ABT-737 exposure. ABT-737 caused death of C91PL T-cells in a concentration-dependent manner (Fig.  8A). As a control and in agreement with a previous report (70), Jurkat T-cells were poorly sensitive to ABT-737 (data not shown). We then combined treatment with ABT-737 and Bfl-1 shRNA. Interestingly, Bfl-1 knockdown resulted in complete sensitization of C91PL T-cells to ABT-737 treatment (Fig. 8B). Such cell death was not observed in C91PL cells transduced with Bfl-1 shRNA plus DMSO or with Ctl shRNA plus ABT-737. As a negative control, Bfl-1 knockdown did not significantly induce cell death of ABT-737-treated Jurkat T-cell line (data not shown).
Thus, our data demonstrate for the first time that Bfl-1 expression confers resistance of the HTLV-1-infected C91PL T-cell line to ABT-737 treatment, and the data suggest that the concurrent up-regulation of both Bfl-1 and Bcl-x L can synergize to induce survival of HTLV-1-infected T-cell lines.

HTLV-1-infected T-cell Survival Is Compromised by Combination of Bfl-1 Knockdown and Etoposide
Treatment-As acute aggressive ATLL cells were resistant to chemotherapeutic agents (71), we further determined whether expression of Bfl-1 could be associated with resistance to drugs such as etoposide. We first explored viability of the C91PL T-cell line after exposure to etoposide for 48 h. Etoposide caused death of C91PL T-cells in a concentration-dependent manner (Fig. 8C), with a maximum of 60% of dead cells, even for concentrations higher than 1.5 M (data not shown). We next assessed whether Bfl-1 silencing affected etoposide-induced apoptosis of HTLV-1-  Fig. 8D, Bfl-1 shRNA, but not Ctl shRNA, efficiently increased sensitivity to etoposide. Such an effect is not observed in the presence of DMSO (Fig. 8D). Altogether, these data demonstrate a role for Bfl-1 in resistance to etoposide-induced apoptosis of HTLV-1-infected T-cells.

DISCUSSION
The role of the antiapoptotic Bfl-1 and Bcl-x L proteins in tumor survival and chemoresistance to current drugs has been documented (72). ATLL development is a multistep oncogenic process, which affects among others cellular functions, proliferation, and cell survival. The outcome of aggressive ATLL remains very poor because HTLV-1-infected cells are resistant to most conventional chemotherapy. Understanding the impact of antiapoptotic proteins of the Bcl-2 family on HTLV-1-infected T-cell survival may provide clues for the development of new therapeutic strategies. Although the expression of Bcl-2 and Bcl-x L antiapoptotic proteins has been extensively documented in HTLV-1-infected T-cell lines and in ATLL cells, the role of Bfl-1 has never been described. Here, we showed the following: (i) HTLV-1-infected T-cells express Bfl-1 protein, whereas uninfected leukemia/lymphoma T-cells do not; (ii) the regulatory viral Tax protein induces Bfl-1 expression through both NF-B and AP-1 pathways, whereas the HBZ protein modulates AP-1-induced Bfl-1 expression; and (iii) Bfl-1 and Bcl-x L , but not Bcl-2, are involved in the survival of HTLV-1-infected T-cells.
The first link between HTLV-1 infection and the deregulation of bfl-1 expression was revealed by gene expression profile experiments indicating that bfl-1 gene is significantly up-regu-  Results are expressed as mean Ϯ S.D. of three independent experiments. D, C91PL cells were transduced with Bfl-1 or scramble Bfl-1 Ctl shRNA and 5 days later were exposed to different concentrations of etoposide for 2 additional days. Apoptosis was assessed by anti-active caspase-3 Ab staining and measured by flow cytometry. Results are from one representative experiment of two. lated in PHA/IL-2-stimulated HTLV-1-infected human peripheral blood T-cells, compared with stimulated uninfected T-cells (73). Reinforcing this observation, our results, based on the analysis of six HTLV-1-infected and five uninfected T-cell lines, indicate that Bfl-1 expression is restricted to HTLV-1infected T-cell lines. Interestingly, we also observed Bfl-1 protein expression in two acute ATLL cells, but not in two chronic ATLL cells. Such results suggest that Bfl-1 expression might be related to the aggressive status of acute ATLL associated with a poor prognosis and multidrug resistance, although the chronic ATLL is considered as an indolent subtype (71). However, more samples need to be studied to definitively demonstrate overexpression of Bfl-1 in aggressive ATLL cells.

Tax of HTLV-1 Mediates T-cell Survival via
More and more studies point out the growing complexity of the mechanisms regulating viral Tax and HBZ protein expression and their impact in leukemogenesis (see reviews in Refs. 14,21,28). According to our first observation in primary thymocytes (74), we found here that Tax alone up-regulates endogenous bfl-1 in primary mature T-cells, indicating that Bfl-1 expression in HTLV-1-infected T-cells is at least Tax-dependent. Our data show unequivocally that Tax-mediated Bfl-1 expression occurs in part through the NF-B pathway, involving both the canonical and alternative pathways. Supporting the involvement of NF-B in HTLV-1-mediated T-cell survival, we and others observed that Bay11-7082, an inhibitor of the canonical NF-B pathway, partly down-regulates Bcl-x L (75) and Bfl-1 (our work), and partially induced cell death in HTLV-1-infected T-cell lines (data not shown). These observations support the hypothesis that signaling pathways, other than the NF-B pathway, are involved in overexpression of the antiapoptotic Bfl-1 protein. We showed that JunD and c-Jun, but not JunB, potently activate the bfl-1 promoter and stimulate endogenous expression of bfl-1. Jun members are responsible for the expression of genes involved in cell proliferation, survival, and T-cell transformation (76,77). Although Tax is able to activate the expression and to increase the activity of JunD, c-Jun, c-Fos, and Fra-1/2 in vitro (20, 78 -80), only JunD seems to be constitutively activated in ATLL cells (27,78). We report here that JunD or c-Jun synergized with Tax to activate bfl-1 transcription. We observed both c-Jun and JunD binding activities in two sites of the proximal part of the bfl-1 promoter in E12 T-cell clone highly expressing Tax. Altogether, these data demonstrate for the first time that NF-B and AP-1 members mediate Tax-induced Bfl-1 expression. Thus, we propose that both NF-B and AP-1, especially JunD, transcription factors, are involved in Tax-mediated antiapoptotic activity through the up-regulation of the bfl-1 protein. In keeping with such an hypothesis, we observed that knockdown of Tax down-regulates bfl-1 expression and induces death of the E12 T-cell clone expressing both Tax and HBZ (data not shown). These findings are in agreement with previous observations showing that the mouse CTLL-2 T-cells expressing the Tax mutant defective for its ability to activate both NF-B and AP-1 (Tax-M22) were unable to survive in the absence of IL-2, in contrast to cells expressing Tax-WT (79,81).
Although Tax plays an important role in the initiation to leukemogenesis, HBZ seems to rather maintain the tumoral phenotype of ATLL cells. Numerous studies showed that HBZ modulates NF-B and AP-1 pathways to enhance proliferation of HTLV-1-infected T-cells in the late stages of ATLL (for review see Ref. 28). The present data demonstrate that HBZ is unable to trans-activate the bfl-1 gene in T-cells and that it inhibits Tax-mediated bfl-1 trans-activation. Supporting our observation, Zhi et al. (82) reported that HBZ decreases Taxmediated NF-B activation. Zhao et al. (83) indicated a selective suppression of the canonical NF-B pathway by HBZ, which interacts with the p65 subunit of NF-B, thus preventing its DNA binding activity and inducing its degradation through a proteasome-dependent pathway. However, HBZ-mediated repression of the canonical NF-B pathway is not efficient when Tax is overexpressed (19,83,84). HBZ has also been described to modulate gene expression, especially hTERT, depending on its Jun partner (29). Here, we report that ectopic expression of HBZ inhibits c-Jun-mediated bfl-1 gene activation and enhances bfl-1 promoter activity induced by JunD in transfected cells. We also showed that Bfl-1 is highly expressed in T-cell lines, which constitutively express a low level of HBZ and a high level of Tax, supporting the notion that Tax seems to be the major inducer of Bfl-1, compared with HBZ, when Tax is highly expressed. However, we cannot exclude that other viral proteins may regulate Bfl-1 expression. For example, accessory p30 II protein of HTLV-1 is known to modulate apoptotic pathways, by regulating the transcriptional activity of NF-B and AP-1 (85). In agreement with this hypothesis, we found no correlation between the expression of Tax and/or HBZ and that of Bfl-1 in HTLV-1-infected T-cell lines.
Finally, our results directly implicate both Bfl-1 and Bcl-x L in the survival of HTLV-1-infected T-cell lines and support the hypothesis that these two antiapoptotic proteins may be crucial for the extended survival of (pre-)leukemic T-cells. Similarly to what was previously observed by us and others in B lymphomas (47,86), silencing of the Bfl-1 antiapoptotic protein induces significant cell death of HTLV-1-infected T-cells in a caspasedependent manner. In agreement with the model proposed by Certo et al. (87), our observation suggests that inhibition of Bfl-1 antiapoptotic protein by itself, either by down-regulating its expression or through interaction with BH-3-only proteins or BH3 mimetics, may be sufficient to provoke cell apoptosis.
Drugs used in ATLL treatment, such as the NF-B inhibitor Bay11-7082, suppress Bfl-1 and Bcl-x L expression and induce apoptosis of HTLV-1-infected T-cell lines (data not shown). Other drugs, such as arsenic trioxide, have also been shown to induce down-regulation of Bcl-x L , but not that of Bcl-2, and apoptosis in HTLV-1-infected T-cell lines and in ex vivo ATLL cells (88). Apoptosis of tumor cells can be triggered by small compounds that mimic BH3-only proteins and target antiapoptotic Bcl-2 proteins (70). However, ABT-737, which targets Bcl-2, Bcl-x L and Bcl-w (66), does not induce an efficient in vitro apoptosis of ATLL cells expressing at least one member of the Bcl-2 family, such as Bcl-2, Bcl-x L , Bcl-w, and Mcl-1 (89). The low sensitivity of ATLL cells to ABT-737-induced apoptosis may indeed be explained by overexpression of Mcl-1, but also Bfl-1, because both proteins are known to be involved in ABT-737 resistance (90 -93). In agreement with such an hypothesis, we demonstrate that down-regulation of Bfl-1 expression by shRNA strategy sensitizes the HTLV-1-infected C91PL T-cell

Tax of HTLV-1 Mediates T-cell Survival via Bfl-1 and Bcl-x L
line to ABT-737 treatment. Taken together, our results implicate Bfl-1 and Bcl-x L as crucial regulators of the extended survival of leukemic T-cells in the context of HTLV-1 infection. A previous report indicates that anti-apoptotic activity of Bcl-x L contributes to chemotherapeutic drug resistance of acute ATLL cells (89). To date, the role of Bfl-1 in resistance to conventional chemotherapeutic agents has only been demonstrated in B-cell malignancies (47,86,94). In this study, we show for the first time that Bfl-1 silencing increases etoposide-induced caspase-dependent apoptosis in HTLV-1-infected T-cells. But, etoposide treatment in combination with ABT-737 (89) or with Bfl-1 silencing (our work) shows no synergistic effect, suggesting that both Bfl-1 and Bcl-x L may be simultaneously targeted to restore strong chemosensitivity of HTLV-1infected T-cells. Taken together, these data indicate that these two antiapoptotic proteins may be considered as potential therapeutic targets in T-cell malignancy associated with HTLV-1.