Human T-Cell Leukemia Virus Type 3 (HTLV-3) and HTLV-4 Antisense-Transcript-Encoded Proteins Interact and Transactivate Jun Family-Dependent Transcription via Their Atypical bZIP Motif

ABSTRACT

Human T-cell leukemia virus types 3 and 4 (HTLV-3 and HTLV-4) are recently isolated retroviruses. We have previously characterized HTLV-3- and HTLV-4-encoded antisense genes, termed APH-3 and APH-4, respectively, which, in contrast to HBZ, the HTLV-1 homologue, do not contain a typical bZIP domain (M. Larocque É Halin, S. Landry, S. J. Marriott, W. M. Switzer, and B. Barbeau, J. Virol. 85:12673–12685, 2011, doi:10.1128/JVI.05296-11). As HBZ differentially modulates the transactivation potential of various Jun family members, the effect of APH-3 and APH-4 on JunD-, c-Jun-, and JunB-mediated transcriptional activation was investigated. We first showed that APH-3 and APH-4 upregulated the transactivation potential of all tested Jun family members. Using an human telomerase catalytic subunit (hTERT) promoter construct, our results also highlighted that, unlike HBZ, which solely modulates hTERT expression via JunD, both APH-3 and APH-4 acted positively on the transactivation of the hTERT promoter mediated by tested Jun factors. Coimmunoprecipitation experiments demonstrated that these Jun proteins interacted with APH-3 and APH-4. Although no activation domain was identified for APH proteins, the activation domain of c-Jun was very important in the observed upregulation of its activation potential. We further showed that APH-3 and APH-4 required their putative bZIP-like domains and corresponding leucine residues for interaction and modulation of the transactivation potential of Jun factors. Our results demonstrate that HTLV-encoded antisense proteins behave differently, and that the bZIP-like domains of both APH-3 and APH-4 have retained their interaction potential for Jun members. These studies are important in assessing the differences between HBZ and other antisense proteins, which might further contribute to determining the role of HBZ in HTLV-1-associated diseases.

IMPORTANCE HBZ, the antisense transcript-encoded protein from HTLV-1, is now well recognized as a potential factor for adult T-cell leukemia/lymphoma development. In order to better appreciate the mechanism of action of HBZ, comparison to antisense proteins from other HTLV viruses is important. Little is known in relation to the seemingly nonpathogenic HTLV-3 and HTLV-4 viruses, and studies of their antisense proteins are limited to our previously reported study (M. Larocque É Halin, S. Landry, S. J. Marriott, W. M. Switzer, and B. Barbeau, J. Virol. 85:12673–12685, 2011, doi:10.1128/JVI.05296-11). Here, we demonstrate that Jun transcription factors are differently affected by APH-3 and APH-4 compared to HBZ. These intriguing findings suggest that these proteins act differently on viral replication but also on cellular gene expression, and that highlighting their differences of action might lead to important information allowing us to understand the link between HTLV-1 HBZ and ATL in infected individuals.

INTRODUCTION

Human T-cell leukemia virus type 1 (HTLV-1) is the etiological agent of adult T-cell leukemia/lymphoma (ATL) and HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP) (1–4). Although a significant number of individuals are estimated to be infected with HTLV-1 globally, less than 5% of infected individuals eventually develop ATL (5). HTLV-2 is genetically related to HTLV-1 and shares most viral genes with it, but it differs by having a preferential tropism toward CD8⁺ T cells compared to a CD4⁺ T-cell tropism for HTLV-1 (6). However, recent studies suggest this preference is not clearly present during early infection (7). Furthermore, HTLV-2 infection has a clinical presentation distinct from that of HTLV-1 and has been linked only to HAM-like pathologies and not to leukemia (8–10). Recently, two new HTLV viruses, termed HTLV-3 and HTLV-4, emerged in nonhuman primate hunters from Cameroon (11–13). Although HTLV-3 presents certain similarities to HTLV-1, such as a Tax protein with common functional features, diseases have not been reported in HTLV-3-infected individuals (nor individuals infected by HTLV-4) thus far. Further epidemiologic studies are needed to determine the public health significance of these emerging viral infections (14–16).

The mechanism by which HTLV-1 induces ATL is not well understood but has been associated with its transactivator protein (Tax) (17). Tax is essential for HTLV-1 replication by its capacity to activate transcription factors acting on the long terminal repeat (LTR) activity but also can transform human primary CD4⁺ T cells and induce different tumors in mice (18–21). Cellular transformation depends on the ability of Tax to interfere with cell cycle modulation and apoptosis, to induce genetic instability, and to regulate multiple transcription factors, including CREB, SRF, and AP-1 (5, 22–24). However, Tax often is not expressed in cells from ATL patients in part due to epigenetic changes or genetic inactivation; thus, it is dispensable at least during late stages of leukemogenesis (25). Selective pressure mediated by a strong anti-Tax immune response might also downregulate Tax expression in ATL cells (26).

HTLV-1 also encodes a protein expressed from the antisense strand and termed HBZ (HTLV-1 bZIP). HBZ exists as two different isoforms derived from unspliced or spliced transcripts (27–30). Unlike the tax gene, the abundant spliced HBZ transcript is detected in most ATL cell lines and peripheral blood lymphocytes (PBLs) from infected patients (28–30). HBZ has been linked to cell proliferation, induction of T-cell lymphomas, and systemic inflammation in transgenic mice, and its expression correlates with high proviral load and severity of HAM symptoms (31–33). HBZ interacts with several transcription factors through its bZIP motif, including the bZIP-containing Jun factors, which block basal and Tax-induced HTLV-1 transcription (23, 24, 27, 34–37).

The AP-1 family of transcription factors belongs to the bZIP class of DNA binding proteins. Protein-protein interactions are dependent on the ZIP region, which is characterized by leucine heptad repeats forming an α-helix-based coiled-coil structure (38–40). Most AP-1 homo- or heterodimers bind to TPA-responsive elements (TRE) (41, 42), while the CREB/c-Jun heterodimer binds cyclic AMP (cAMP)-responsive elements (CRE) (43) through a basic-rich region positioned upstream of the leucine zipper (LZ) (38, 40). AP-1 dimers tightly modulate transcription of various genes involved in cellular processes, such as transformation, proliferation, survival, and apoptosis (44, 45). Interestingly, HBZ has been shown to interact with several Jun family members, inactivating them through possible sequestration in transcriptionally inactive nuclear bodies or by degradation (34, 35, 37, 46, 47). In contrast, former studies have shown that HBZ also may interact with and augment the JunD transactivation potential and increase expression of both HBZ and the human telomerase catalytic subunit (hTERT) (48, 49).

Recently, we reported that HTLV-2 also produces an antisense protein, termed APH-2 (50). In addition, we have identified and partially characterized similar antisense proteins in HTLV-3 and HTLV-4, termed APH-3 and APH-4 (51). Although these three new viral proteins lack a classical bZIP domain, they all share the capacity to act upon Tax-mediated HTLV LTR activation. Nonetheless, more recent studies have shown that HBZ and APH-2 utilize different mechanisms for downregulation of HTLV transcription, raising questions about similar differences in APH-3 and APH-4 (10, 50, 52, 53). Here, we investigated the potential modulation of Jun family members by APH-3 and APH-4 and demonstrate that, as opposed to HBZ, both viral proteins augment the transcriptional activation potential of JunD, c-Jun, and JunB through their bZIP-like domain.

MATERIALS AND METHODS

Cell lines.

293T, COS-7, and HeLa cells were cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (PAA Laboratories Inc., Toronto, Canada). The CEM T-cell line was maintained in RPMI 1640 medium supplemented with 10% FBS. All cell lines were cultured in a 5% CO₂ atmosphere at 37°C.

Plasmid constructs.

HBZ-SP1-, APH-3-, and APH-4-expressing vectors have been previously described and express green fluorescent protein (GFP) chimeric or Myc-tagged proteins (pHBZ-SP1-Myc, pMycAPH-3, pMycAPH-4, pAPH-3-GFP, and pAPH-4-GFP) (51). Deletion mutants were generated from pMycAPH-3, pAPH-3-GFP, pMycAPH-4, or pAPH-4-GFP (51) using a Phusion site-directed mutagenesis kit according to the manufacturer's instructions. The following forward and reverse primers were used to generate the APH-3 and APH-4 deletion mutants: 5′-GAG GAA TAC TTT TGG TAT ATG GAC TGT T-3′ and 5′-GTC TAT ACA GTC CAG TTG CTG TCG-3′ for APH-3Δ157-189, 5′-GGA GAA TTA TTT CGG TTA TCA GCT GG-3′ and 5′-GTC CAT GCA CTT CCG TTG TTG ATG-3′ for APH-4Δ103-135, 5′-GAG GGC TAT ATA GAC TTT TTG GAG-3′ and 5′-GTC TAT ACA GTC CAG TTG CTG TCG-3 for APH-3Δ157-167, 5′-CAG CAG CTA CGC GCA GGC T-3′ and 5′-TAG TAA GTC ACA AAA CCC CTC GAA C-3′ for APH-3Δ168-178, 5′-GAG GAA TAC TTT TGG TAT ATG GAC TGT T-3′ and 5′-GCT CTC CCG CTC CAA AAA GTC TA-3′ for APH-3Δ179-189, 5′-CAA GGC TAT ATT GAT GTA TTA GAG AAG GAA-3′ and 5′-GTC CAT GCA CTT CCG TTG TTG ATG-3′ for APH-4Δ103-113, 5′-AAG GAT CTG GAG GAG GAA TGT AAC C-3′ and 5′-CAA CAA GTC ACA ATA CCC CTC AAA CC-3′ for APH-4Δ114-124, and 5′-GGA GAA TTA TTT CGG TTA TCA GCT GG-3′ and 5′-CCT TTC CTT CTC TAA TAC ATC AAT ATA GC-3′ for APH-4Δ125-135. Point mutation constructs of APH-3 and APH-4 also were generated using the Phusion site-directed mutagenesis kit. The forward 5′-GTA GAG GGC TAT ATA GAC TTT GTG GA-3′ and reverse 5′-TAA GTC ACA AAA CCC CTC GAA CCC-3′ primers and forward 5′-GTG CAA GGC TAT ATT GAT GTA GTA GAG A-3′ and reverse 5′-CAA GTC ACA ATA CCC CTC AAA CCC-3′ primers were used to generate pMycAPH-3L₁₆₇L₁₇₄-VV and pMycAPH-4L₁₁₃L₁₂₀-VV, respectively. The reporter gene construct in which the collagenase promoter (bearing a single AP-1-binding site) is cloned upstream of the luciferase reporter gene was previously described (46). The pAP-1-Luc vector (Stratagene) contains seven AP-1-binding sites upstream of a minimal promoter and the luciferase reporter gene. The wild-type (WT) c-Fos-, JunB-, c-Jun-, and JunD-expressing vectors pcDNAc-Fos, pCMVJunDFlag, pcDNAJunB, and pcDNAc-Jun and the dominant-negative c-Jun pCMV-TAM67 (deleted of its transactivating domain [amino acids 1 to 122]) mutant expression vector all have been described previously (54). The pGL3-378 hTERT construct contains the human telomerase RT promoter region (from −378 to +1) upstream of the luciferase reporter gene (55). The pRcActin-LacZ vector expresses the β-galactosidase gene under the control of the human β-actin promoter. pCMV-Hyg expresses the hygromycin resistance gene. The pRL-TK vector expresses the renilla luciferase reporter gene (48).

Immunofluorescence microscopy analysis.

COS-7 cells were transfected with pAPH-3-GFP, pAPH-3Δ157-189-GFP, pAPH-4-GFP, pAPH-4Δ103-135-GFP, or the control empty vector, peGFP-N1. Thirty-six hours after transfection, cells were washed three times with phosphate-buffered saline (PBS) and incubated for 5 min at 37°C with 1 μg/ml Hoechst 33342 reagent (Invitrogen). Cells then were washed three times with PBS, replenished with supplemented DMEM, and observed at room temperature using a Plan Fluor 40× objective with a numerical aperture (NA) of 0.60 on a Nikon A1 laser-scanning confocal microscope (Nikon Canada, Mississauga, Canada).

Transfection and luciferase assay.

293T cells (1 × 10⁵) were transiently cotransfected with the collagenase luciferase reporter construct (0.4 μg) and either pCMVJunDFlag, pcDNAJunB, or pcDNAc-Jun (0.2 μg), together with 0.4 μg of expression vectors for HBZ-SP1-Myc, wild-type or mutated MycAPH-3 or MycAPH-4, c-Fos, or the empty vector, pcDNA3.1, along with pRcActin-lacZ (0.2 μg), using the Lipofectamine 2000 reagent. Forty-eight hours posttransfection, cells were lysed and luciferase activity was determined with an MLX microplate luminometer (Dynex Technologies), as previously described (56). Each sample was normalized by β-galactosidase (β-Gal) activity for transfection efficiency using a Galacto-Light kit (Applied Biosystems, Bedford, MA) according to the manufacturer's protocol. For each sample, luciferase activity is presented as fold induction of the normalized luciferase activity (relative light units [RLU]/β-Gal) divided by the corresponding normalized luciferase activity from cell samples transfected with the luciferase reporter construct and empty expression vectors, and this value represents the calculated means ± standard errors of the means (SEM) from six transfected samples. CEM cells (1 × 10⁷) were cotransfected with the collagenase luciferase reporter construct (1 μg), pcDNAc-Jun (1 μg), and 2 μg of pHBZ-SP1-Myc, pMycAPH-2, pMycAPH-3, pMycAPH-4, or pcDNA3.1, as well as with pRcActin-lacZ, using the Gene Pulser Xcell electroporation system (1,200 μF, 250 V; Bio-Rad, Hercules, CA), as previously described (57). HeLa cells were cotransfected using Polyfect (Qiagen) according to the manufacturer's instructions with 0.1 μg pGL3-378 hTERT, 0.025 μg pCMVJunD, 0.2 μg pcDNAJunB or pcDNAc-Jun, and 0.2 μg pHBZ-SP1-Myc, pMycAPH-3, pMycAPH-4, or pcDNA3.1. Each sample was cotransfected with a renilla luciferase expression vector (pRL-TK; Promega) for normalization of transfection efficiency. Forty-eight hours posttransfection, activity was measured using the dual-luciferase reporter assay (Promega) according to the manufacturer's instructions. Luciferase activity is presented as fold induction of the value of the reporter plasmid alone.

Generation of stable cell line and transfection.

293T cells (1 × 10⁵) were stably transfected with the collagenase-luc vector (or pcDNA3.1 as a control) along with pCMV-Hyg (ratio of 10:1) using the Lipofectamine 2000 reagent. Cells were selected in the presence of 400 ng/μl hygromycin B, and resulting clones were pooled. Stable cell lines then were transiently cotransfected with pCMVJunDFlag, pcDNAJunB, or pcDNAc-Jun together with pHBZ-SP1-Myc, pMycAPH-3, pMycAPH-4, or the empty vector, pcDNA3.1, using the Lipofectamine 2000 reagent. Forty-eight hours posttransfection, cells were lysed and luciferase activity was quantified.

Yeast two-hybrid assay.

The wild type (WT) or bZIP-deleted Jun cDNA or c-Jun AD deletion mutants were cloned in frame with the Gal4 activation domain (AD) in the pGAD10 vector. Wild-type APH-4, APH-4Δ109-182, or APH-4Δ145-182 cDNAs were cloned in frame with the Gal4 DNA binding domain of the pGBT9 vector. A yeast two-hybrid assay was performed in the Saccharomyces cerevisiae HF7c reporter strain, which contains the Escherichia coli lacZ gene under the control of three copies of the GAL4-binding sequence. As previously described (35), HF7c cells were cotransformed using the lithium acetate method with the WT pGAD10-based expression vector or mutated JunD, c-Jun, or JunB (or the empty vector) and with pGBT9 expression vectors for WT or mutated APH-4. The experiment was carried out on three independent pools of colonies per transformation as described in the Clontech protocol. The β-galactosidase activity was measured for each sample using o-nitrophenyl-β-d-galactopyranoside as the substrate and then calculated in Miller units (58).

GAL4-based activation domain assay.

APH-3 and APH-4 were cloned in frame into the pBind vector containing the DNA binding domain of GAL4 (pBIND vector; Promega). CEM and 293T cells were cotransfected with pG5-luc, a luciferase reporter plasmid with a minimal TATA box promoter downstream of five GAL4-binding sites, together with pBind APH-2, pBind APH-3, pBind APH-4, pBind HBZ (27), or the empty vector, along with pRcActin-lacZ for transfection efficiency, as described above. Luciferase activity was measured at 48 h posttransfection and is presented as the normalized fold induction compared to luciferase activity from pBind-transfected cells. Data represent the calculated means ± SEM from two transfected samples.

Western blotting.

Forty μg of total cell extract was run on a 12% SDS-PAGE gel and transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, Mississauga, Canada). The blot was blocked in PBS-5% milk and incubated with an anti-myc (1:250) (supernatant of hybridoma purchased from the American Tissue Culture Collection [Manassas, VA]) or anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH; 1:1,000) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) antibody. After several washes, horseradish peroxidase (HRP)-conjugated sheep anti-mouse IgG antibodies (1:5,000) were added to the membranes, which were washed again several times and subsequently incubated with the BM chemiluminescence blotting substrate (POD) (Roche Diagnostic, Indianapolis, IN). Membranes were analyzed with the Fusion FX7 apparatus (Vilber Lourmat, France).

Co-IP experiments.

293T cells were cotransfected with pMycAPH-3 or pMycAPH-4 and pcDNA-JunB, pcDNA-c-Jun, or pCMV-JunD-Flag using the Lipofectamine 2000 reagent. Forty-eight hours posttransfection, cells were lysed with 50 mM Tris-HCl, pH 8, 100 mM NaCl, 1 mM EDTA, and 1% Triton X-100. Immunoprecipitation (IP) was done using Dynabeads protein G (Invitrogen) according to the manufacturer's instructions. Briefly, 40 μl of Dynabeads protein G (Invitrogen) was incubated for 30 min at room temperature with 5 μg of anti-myc antibody. Total cell extracts then were incubated overnight at 4°C with the antibody-bead complex. Bound fractions were eluted with 20 μl of loading buffer and then separated using a 12% SDS-PAGE and transferred to a PVDF membrane (Millipore, Mississauga, Canada). Membranes were blocked in PBS-5% milk or PBS-0.5% casein and incubated with an anti-JunB (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) (1:500), anti-c-Jun (Santa Cruz Biotechnology, Inc.) (1:200), anti-JunD (Santa Cruz Biotechnology, Inc.) (0.4 μg/μl), or anti-myc (dilution, 1:250) antibody. After several washes, membranes were incubated with horseradish peroxidase (HRP)-conjugated mouse IgG Trueblot ultra (eBioscience, San Diego, CA) (1:1000) or HRP-conjugated sheep anti-rabbit IgG antibodies (1:5,000), washed several times, and incubated with the BM chemiluminescence blotting substrate (POD). Membranes were analyzed using the Fusion FX7 device.

Statistical analysis.

Data are shown as means ± SEM and analyzed by one-way analysis of variance (ANOVA) using Tukey's test. Differences were considered statistically significant when P ≤ 0.05.

RESULTS

APH-3 and APH-4 increase the transactivation potential of Jun members.

While HBZ downregulates JunB- and c-Jun-dependent transcription and upregulates transcription dependent on JunD (59–61), it is unknown if APH-3 and APH-4 could similarly affect the Jun family of transcription factors. To test this possibility, 293T cells were transiently cotransfected with a construct expressing the luciferase reporter gene positioned downstream of the collagenase promoter (bearing one AP-1-binding site) and either pCMVJunDFlag, pCMVJunB, or pCMVc-Jun together with pHBZ-SP1-Myc, pMycAPH-3, pMycAPH-4, or the empty vector. In Fig. 1A to C, we confirm that HBZ downregulates the transactivation potential of JunB and c-Jun in the context of the collagenase promoter, while it incremented the activation potential of JunD (46, 61). Surprisingly, both APH-3 and APH-4 further activated the transactivation potential of all three Jun family members, with most differences being statistically significant compared to levels for cells expressing only the Jun factor. Western blot analyses of cellular extracts of transfected cells confirmed that these differences were not attributable solely to variable levels of either Jun factors or APH-3 and APH-4. APH-3 and APH-4, when expressed alone, also positively modulated Jun-mediated transcriptional activation, which suggests an effect of APH-3 and APH-4 on the transactivation potential of endogenous Jun factors. When an APH-2 expression vector was similarly cotransfected with the JunD expression vector in 293T cells, activation of the collagenase promoter mediated by JunD again was induced further, albeit at a lower level than that for APH-3 and APH-4 (data not shown). However, the very low level of detectable APH-2 protein in transfected cells did not allow us to determine if APH-2 was less capable of inducing JunD-dependent activation of the collagenase promoter than APH-3 and APH-4.

FIG 1.

APH-3 and APH-4 activate Jun-dependent transcription of the collagenase promoter. (A to C) pHBZ-SP1-Myc, pMycAPH-3, pMycAPH-4, or the empty vector, pcDNA3.1, was cotransfected in 293T cells along with the expression vector for JunB (A), c-Jun (B), or JunD (C), the collagenase-luciferase construct, and pRcActin-lacZ. Forty-eight h posttransfection, cells were lysed and luciferase activity was measured. Luciferase activity is presented as fold induction ± standard errors of the means (SEM) from six transfected samples, in which the value for cells transfected with the luciferase reporter construct and empty expression vectors was set at 1. These results are representative of four independent experiments. In lower panels, cell lysates analyzed by Western blotting with anti-myc and anti-GAPDH antibodies as well as anti-JunB, anti-c-Jun, or anti-JunD antibody are presented. Arrows on the right side of the images point to specific signals. (D) CEM cells were transiently cotransfected with collagenase-luc, pcDNAc-Jun, pHBZ-SP1-Myc, pMycAPH-2, pMycAPH-3, pMycAPH-4, or the empty vector, pcDNA3.1, along with pRcActin-lacZ (left). On the right, CEM cells were similarly cotransfected with pRcActin-lacZ, collagenase-luc, c-Jun, c-Fos, and/or APH-4 expression vectors. Luciferase activity was measured at 48 h posttransfection. Results were normalized for β-galactosidase activity, and fold induction was calculated by dividing each luciferase activity value by values of cells transfected in the absence of expression vectors. Data represent the mean fold induction from triplicates ± SEM. These results are a representation of three independent experiments (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001).

To further confirm these results, CEM cells also were transfected with c-Jun expression vectors and measured for collagenase promoter activity in the presence of HBZ, APH-2, APH-3, or APH-4 (Fig. 1D, left). Again, APH-2, APH-3, and APH-4 all increased c-Jun-mediated activation of the collagenase promoter, corroborating results from transfected 293T cells. However, due to the high intrinsic instability of APH-2, again we were unable to clearly state how the transactivation potential of this protein compared to those of APH-3 and APH-4. Since APH-4 most strongly increased c-Jun-dependent promoter activation, we next compared its transactivation potential to that of the c-Fos/c-Jun heterodimer. Comparable transactivation potentials of the collagenase promoter (about 300-fold) were obtained in CEM cells cotransfected by either c-Fos or APH-4 and the c-Jun expression vectors (Fig. 1D, right).

Our results demonstrate that APH-3 and APH-4 positively upregulate the activation of the collagenase promoter, and that the extent of this upregulation varies depending on the coexpressed Jun factor.

APH-3 and APH-4 augment Jun activation in the chromatin environment and upstream of a heterologous promoter.

The observed induction of the collagenase promoter by APH-3/APH-4 and Jun family members was further evaluated using 293T cells stably transfected with the collagenase promoter luciferase construct. These 293T cells were transfected with the JunD expression vector and APH-3 or APH-4 expression vectors. As expected, both APH-3 and APH-4 potently augmented luciferase activity in the presence of JunD (Fig. 2A and B). Interestingly, APH-4 alone also induced collagenase promoter activity.

FIG 2.

APH-3 and APH-4 activate Jun-dependent transcription in a heterologous promoter and in the environment of chromatin. (A and B) 293T cells stably transfected with the collagenase-luc construct were cotransfected with pCMVJunDFlag (or the empty vector) along with either pMycAPH-3 (A) or pMycAPH-4 (B). Forty-eight h posttransfection, cells were lysed and luciferase activity was measured. (C) pMycAPH-3, pMycAPH-4, or the empty vector (pcDNA3.1) was cotransfected in 293T cells along with pAP-1-luc, pCMVJunDFlag, and pRcActin-lacZ. Forty-eight h posttransfection, cells were lysed and luciferase activity was measured. Luciferase activity is presented as the calculated fold induction from the normalized luciferase activity of cell samples transfected with the empty expression vector (cotransfected with pAP-1-luc for panel C). These results are a representation of three independent experiments (*, P ≤ 0.05; ***, P ≤ 0.001).

We also conducted a series of transfections with a construct containing the luciferase reporter gene positioned downstream of a minimal promoter and seven copies of the AP-1 consensus sequence. We found that luciferase activity was induced by JunD, with increased induction when either APH-3 or APH-4 was expressed (Fig. 2C). Our results suggest that APH-3 and APH-4 had similar effects on the collagenase promoter in the environment of chromatin and also that induction could be confirmed with AP-1-binding sites positioned next to a minimal promoter.

APH-3 and APH-4 activate the hTERT promoter.

Previous studies reported that HBZ could increase promoter activity of the human telomerase hTERT gene by its ability to complex with JunD and Sp1 (48). Since both APH-3 and APH-4 induced the transactivation potential of the tested Jun family members, we next evaluated whether hTERT promoter activity could be upregulated equally by APH-3 and APH-4. HeLa cells were transfected with a construct expressing the luciferase reporter gene positioned downstream of the hTERT minimal promoter (bearing Sp1- and c-myc-binding sites and no AP-1-binding sites) in combination with either pCMVJunDFlag, pcDNAJunB, or pcDNAc-Jun and with pHBZ-SP1-Myc, pMycAPH-3, pMycAPH-4, or the empty vector (Fig. 3). As expected, HBZ demonstrated little c-Jun- and JunB-modulated hTERT promoter activity but augmented the activation mediated by JunD. In contrast, both APH-3 and APH-4 further induced hTERT promoter activity regardless of the tested Jun family member. The most notable increase again was mediated by APH-4. In this cellular context, Western blot analyses indicated that HBZ, APH-3, and APH-4 showed comparable expression levels and did not alter cytomegalovirus (CMV)-driven Jun expression when JunD expression vectors were transfected (Fig. 3C). These results illustrate that both APH-3 and APH-4 strongly modulate the transactivation potential of Jun members in the context of the hTERT promoter.

FIG 3.

APH-3 and APH-4 augment hTERT promoter activation by Jun factors. HeLa cells were transiently cotransfected with the pGL3-378 reporter plasmid (containing the hTERT promoter) along with pHBZ-SP1-Myc, pMycAPH-3, pMycAPH-4, or pcDNA3.1 and JunB (A), c-Jun (B), or JunD (C) expression vector. The pRL-TK vector also was cotransfected for normalization of transfection efficiency. Cells were lysed 48 h posttransfection, and luciferase activity was measured and normalized for renilla luciferase activity (pRL-TK). Results are presented as the calculated fold induction from the normalized luciferase activity of cell samples transfected with the reporter construct and the empty expression vector. These results are representative of three independent experiments. In the lower panels, lysates of transfected cells were analyzed by Western blotting with anti-myc or anti-α-actin antibodies.

APH-3 and APH-4 interact with c-Jun, JunB, and JunD.

As our results clearly suggested that APH-3 and APH-4 functionally modulated Jun transcription factors, we next verified if APH-3 and APH-4 could interact directly with JunB, c-Jun, and JunD. 293T cells were cotransfected with pcDNAJunB, pcDNAc-Jun, or pCMVJunDFlag expression vector with either pMycAPH-3 or pMycAPH-4 (Fig. 4). In 293T cells cotransfected with the JunB expression vector, immunoprecipitation of APH-3 and APH-4 with an anti-myc antibody led to detection of JunB, indicating that both proteins indeed were interacting with JunB (Fig. 4A). Similar experiments using c-Jun and JunD expression vectors transfected in 293T cells with APH-3 or APH-4 expression vectors also showed potential interactions between c-Jun or JunD and either antisense protein, as depicted in Fig. 4B and C. Thus, our results suggest that both APH-3 and APH-4 were interacting with all three Jun family members.

FIG 4.

APH-3 and APH-4 interact with each Jun family member. 293T cells were cotransfected with pMycAPH-3, pMycAPH-4, or pcDNA3.1 along with expression vectors for either JunB (A), c-Jun (B), or Flag-tagged JunD (C). Forty hours posttransfection, cells were lysed and lysates were used for immunoprecipitation (IP) with an anti-myc antibody. Immunoprecipitates and cell extracts were analyzed in parallel by Western blotting using an anti-JunB, anti-c-Jun, anti-Flag, or anti-myc antibody. IB, immunoblot.

Lack of an AD for APH-3 and APH-4.

As HBZ contains an AD at its NH₂ end (27, 46), we were interested in searching for a potential activation domain in both APH-3 and APH-4. To better understand the mechanism behind the activating capacity of APH-3 and APH-4 on the Jun transactivating potential, we first generated deletion mutants of the amino-terminal region of these two proteins (Δ1-89 and Δ1-36 for APH-3 and APH-4, respectively) corresponding to regions showing sequence similarity to the HBZ AD (Fig. 5A). When each mutant was transfected with either c-Jun or JunD expression vector and the collagenase promoter-luciferase construct, measurement of luciferase activity showed no significant differences between WT and mutant APH proteins (Fig. 5B and C). To further test for the presence of an activation domain in APH-3 and APH-4, expression vectors of chimeric proteins containing the entire open reading frame (ORF) of all four antisense proteins, i.e., HBZ, APH-2, APH-3, and APH-4, fused to the GAL-4-binding domain were transfected with the pG5luc reporter vector containing five GAL4-binding sites adjacent to a minimal promoter in 293T or CEM cells (Fig. 5D and E). Confirming the results described above, chimeric APH-3 and APH-4 proteins were not efficient in inducing luciferase activity, unlike the chimeric HBZ protein, which showed over 70- and 140-fold induction in 293T and CEM cells, respectively.

FIG 5.

Absence of an activation domain in the N-terminal region of APH-3 and APH-4. (A to C) Expression vectors for APH-3 and APH-4 mutants deleted at their amino-terminal end (pMycAPH-3Δ1-89 and pMycAPH-4Δ1-36) were generated (A) and cotransfected in 293T cells with collagenase-luc, pcDNAc-Jun (B), or pCMVJunDFlag (C), along with pRcActin-lacZ. As a control, cotransfections also were conducted in parallel with wild-type APH-3 or APH-4 expression vector or the empty vector. At 48 h posttransfection, cells were lysed and luciferase activity was measured. Luciferase activity is presented as fold induction ± SEM, in which the value for cells transfected with the luciferase reporter construct and empty expression vectors was set at 1. LZ, leucine zipper. (D and E) 293T (D) and CEM (E) cells were transfected with pG5-luc, pBind HBZ, pBind APH-2, pBind APH-3, pBind APH-4, or the pBind empty vector, along with pRcActin-lacZ. Cell lysates were evaluated for luciferase activity and normalized for β-galactosidase activity. Results are presented as fold induction over pBind-transfected cells and are presented as means ± SEM. (F) CEM cells were cotransfected with collagenase-luc, pcDNAc-Jun, or pcDNAc-JunTAM67, along with pRcActin-lacZ and pMycAPH-4, pMycAPH-4Δ1-36, pMycAPH-4Δ1-80, or the empty vector, pcDNA3.1. Cell lysates were evaluated for luciferase activity and normalized for β-galactosidase activity. Results are presented as fold induction over cells transfected with the reporter construct and empty expression vectors and are presented as the means from triplicates ± SEM.

Since an activation domain was not identified for APH-3 and APH-4 in these experiments, we next tested if Jun members' ADs were involved in the observed APH/Jun-dependent induction of collagenase promoter activity. To test this, a dominant-negative c-Jun mutant deleted of its activation domain, termed TAM67, was cotransfected along with expression vectors for the APH-4 WT or a version deleted of its amino-terminal region (APH-4Δ1-36) or further deleted up to the start of the bZIP domain (APH-4Δ1-80). As seen in Fig. 5F, upon expression of APH-4 in transfected CEM cells, c-Jun strongly induced luciferase expression. In contrast, expression of the dominant-negative c-JunTAM67 strongly hampered APH-4-dependent upregulation of its transactivation potential. Interestingly, although the APH-4Δ1-36 mutant was only modestly affected in its capacity to act upon c-Jun transactivation potential, the APH-4Δ1-80 mutant was more severely affected.

Collectively, these results indicate that APH-3 and APH-4 do not contain an HBZ-like AD and that the observed increase in the transactivation potential of Jun family members was dependent on the Jun AD.

Atypical bZIP motifs of APH-3 and APH-4 are required for their modulation of Jun transactivation potential.

APH-3 and APH-4 both possess a bZIP-like domain showing irregularly spaced leucines or isoleucines, which are not typical of leucine zipper (LZ) motifs. To determine if these atypical LZs affect Jun transactivation, we generated LZ-deleted APH-3 and APH-4 mutants (MycAPH-3Δ157-189 and pMycAPH-4Δ103-135) and assessed if these mutants acted differently on Jun family members (Fig. 6A). Upon transfection in 293T cells, similar subcellular localization and expression levels were noted for both APH-3 and APH-4 mutants and their WT versions, as determined by Western blot analyses and confocal microscopy (Fig. 6B and F). Mutated or WT APH-3 and APH-4 proteins next were coexpressed in 293T cells along with the Jun family members, and their transactivation potential on the cotransfected collagenase-luc vector was determined. As demonstrated in Fig. 6C to E, LZ-deleted APH-3 and APH-4 lost significant hyperactivating potential over Jun family members in the context of the collagenase promoter construct. We next tested if these LZ deletions affected their ability to interact with Jun family members. IP experiments were conducted in 293T cells cotransfected with LZ-deficient APH-3 and APH-4 mutants (as well as WT expression vectors) and with the c-Jun expression vector (Fig. 6G). As expected, when WT APH-3 or APH-4 was immunoprecipitated, c-Jun was detected following Western blot analyses. In contrast, immunoprecipitation of LZ-deleted APH-3 and APH-4 mutants led to a severe reduction in the abundance of c-Jun, confirming a decreased binding efficiency of the deletion mutants to c-Jun.

FIG 6.

Deletion of the APH-3 and APH-4 atypical leucine zipper (LZ) region abolishes the increased transactivation potential of JunB, c-Jun, and Jun-D. (A) Expression vectors for APH-3 and APH-4 mutants deleted in their LZ domain (pMycAPH-3Δ157-189 and pMycAPH-4Δ103-135) were generated and cotransfected in 293T cells. (B) Cell lysates were prepared 48 h posttransfection from 293T cells transfected with pMycAPH-3, MycAPH-3Δ157-189, pMycAPH-4, pMycAPH-4Δ103-135, and pcDNA3.1. Western blot analysis was performed using anti-myc and anti-GAPDH antibodies. (C to E) 293T cells were cotransfected with wild-type (WT) APH-3 and APH-4 or LZ deletion mutants (pMycAPH-3Δ157-189 and pMycAPH-4Δ103-135) together with the collagenase-luc construct, pcDNAJunB (C), pcDNAc-Jun (D), or pCMVJunDFlag (E), along with pRcActin-lacZ. For controls, cotransfections were also conducted in parallel with WT APH-3 or APH-4 expression vector or the empty vector. Cells were lysed and luciferase activity was measured at 48 h posttransfection. Results were normalized by β-galactosidase activity, and luciferase activity is presented as the fold induction ± SEM from six transfected samples, in which the value for cells transfected with the luciferase reporter construct and empty expression vectors was set at 1. (F) COS-7 cells were transfected with pAPH-3-GFP, pAPH-3Δ157-189-GFP, pAPH-4-GFP, pAPH-4Δ103-135-GFP, or peGFP-N1 (control [CTL]). Cell nuclei were labeled with Hoechst (blue) and visualized by confocal microscopy at 48 h posttransfection. GFP signals (green) are presented at the top, and merge signals are shown at the bottom. (G) Cell lysates were prepared 48 h posttransfection from 293T cells cotransfected with pMycAPH-3, MycAPH-3Δ157-189, pMycAPH-4, pMycAPH-4Δ103-135, or pcDNA3.1, along with pcDNAc-Jun. Cell lysates were immunoprecipitated using the anti-myc antibody and further analyzed in comparison to total cellular extracts by Western blotting with anti-c-Jun and anti-myc antibodies (*, P ≤ 0.05; ***, P ≤ 0.001).

To further analyze the region required for the modulatory function of APH-3 and APH-4 on Jun transactivation potential, smaller deletion mutants within the LZ region (APH-3Δ157-167, APH-3Δ168-178, and APH-3Δ179-189 or APH-4Δ103-113, APH-4Δ114-124, and APH-4Δ125-135) were generated and tested by transfection in 293T cells along with Jun expression vectors and the collagenase promoter construct (Fig. 7). Regardless of the Jun family member, all deleted mutants were strongly affected in their activating potential of the collagenase promoter, indicating that the entire LZ region in APH-3 and APH-4 was essential for optimal collagenase promoter activation (Fig. 7B to D). Western blot analyses confirmed that these mutants were expressed at levels equivalent to those of their WT versions (Fig. 7E).

FIG 7.

Partial deletion of the atypical LZ region of APH-3 and APH-4 inhibits the increased transactivation potential of JunB, c-Jun, and Jun-D. (A to D) Expression vectors for APH-3 and APH-4 mutants partially deleted in their LZ domain (pMycAPH-3Δ157-167, pMycAPH-3Δ168-178, pMycAPH-3Δ179-189, pMycAPH-4Δ103-113, pMycAPH-4Δ114-124, and pMycAPH-4Δ125-135) were generated (A) and cotransfected in 293T cells with the collagenase-luc construct, pcDNAJunB (B), pcDNAc-Jun (C), or pCMVJunDFlag (D), along with pRcActin-lacZ. For controls, cotransfections were also conducted in parallel with wild-type APH-3 or APH-4 expression vector or the empty vector. Cells were lysed and luciferase activity was measured at 48 h posttransfection. Luciferase activity is presented as fold induction ± SEM from six transfected samples, in which the value for cells transfected with the luciferase reporter construct and empty expression vectors was at 1. (E) Cell lysates were prepared 48 h posttransfection from 293T cells transfected with pMycAPH-3, pMycAPH-4, pMycAPH-3Δ157-167, pMycAPH-3Δ168-178, pMycAPH-3Δ179-189, pMycAPH-4Δ103-113, pMycAPH-4Δ114-124, pMycAPH-4Δ125-135, and pcDNA3.1. Western blot analysis was performed using anti-myc and anti-GAPDH antibodies (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001).

These results show that the atypical bZIP domain in APH-3 and APH-4 is important for the upregulatory function of both proteins on Jun-mediated collagenase promoter activation, and that it also mediates the interaction with Jun transcription factors.

Implication of leucine residues in the APH-3 and APH-4 LZ domain in the upregulation of Jun transactivation potential.

Through in silico analyses, the position of certain leucines in the atypical LZ domain of APH-3 and APH-4 suggest the interaction of these antisense proteins with other bZIP factors (Fig. 8A). To clearly define these residues in the modulatory function of APH-3 and APH-4 on Jun factors, two leucines similarly positioned in the various antisense proteins in the predicted atypical LZ domain were mutated to valine residues in both APH-3 and APH-4. When expressed in transfected 293T cells, both mutants showed expression levels comparable to those of similarly transfected WT versions (Fig. 8E). Importantly, upon cotransfection with Jun expression vectors and the collagenase promoter reporter construct, mutated APH-3L₁₆₇L₁₇₄-VV and APH-4L₁₁₃L₁₂₀-VV demonstrated a loss of Jun-mediated promoter activation (Fig. 8B to D).

FIG 8.

Leucine residues in the APH-3 and APH-4 atypical LZ region are important for the upregulation of Jun transactivation potential. (A to D) Expression vectors for APH-3 and APH-4 mutated at two leucine residues (pMycAPH-3L₁₆₇L₁₇₄-VV and pMycAPH-4L₁₁₃L₁₂₀-VV) were generated (A) and cotransfected in 293T cells with the collagenase-luc construct, pcDNAJunB (B), pcDNAc-Jun (C), or pCMVJunDFlag (D), along with pRcActin-lacZ. As a control, cotransfections were also conducted in parallel with wild-type (WT) APH-3 or APH-4 expression vector or the empty vector. Cells were lysed and luciferase activity was measured at 48 h posttransfection. Luciferase activity is presented as fold induction ± SEM from six transfected samples in which the value for cells transfected with the luciferase reporter construct and empty expression vectors was set at 1. (E) Cell lysates were prepared 48 h posttransfection from 293T cells transfected with pMycAPH-3, pMycAPH-4, pMycAPH-3L₁₆₇L₁₇₄-VV, pMycAPH-4L₁₁₃L₁₂₀-VV, or pcDNA3.1. Western blot analysis was performed using anti-myc antibodies. (F) pGAD10-based vectors expressing chimeric proteins harboring the Gal4 activation domain fused to wild-type Jun or Jun mutants were cotransformed in the Saccharomyces cerevisiae HF7c reporter strain along with expression vectors for WT or mutant APH-4 (APH-4 Δ109-182, APH-4 Δ145-182, and APH-4L₁₁₃L₁₂₀-VV) fused to the Gal4 DNA-binding domain. Pools of three independent colonies were tested for β-galactosidase activity, and results are presented as mean values ± SEM. These results are representative of three independent experiments (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001).

To further address the importance of the bZIP domain for the interaction of the two APH proteins with Jun family members, a yeast two-hybrid analysis was performed (Fig. 8F). APH-4 was used in these experiments, as it provided the highest induction capacity on Jun-induced promoter activation. WT or APH-4 mutants (APH-4 Δ109-182, APH-4 Δ145-182, and APH-4L₁₁₃L₁₂₀-VV) fused to the GAL-4-binding domain were cotransformed in the HF7c yeast strains along with expression vectors of WT or bZIP-deleted c-Jun, JunB, or JunD fused to the GAL4 activation domain. Upon measurement of β-galactosidase activity in transformed yeast cells, we first showed that lack of the bZIP domain in JunB, c-Jun, and JunD importantly affected the interaction of APH-4 with these factors. Conversely, the loss of the bZIP domain in APH-4 (APH-4 Δ109-182) and mutation of the two selected leucine residues in the previous experiments resulted in a significant reduction in the interaction between c-Jun and APH-4. However, the APH-4 double point mutant had a lesser impact on APH-4/c-Jun interaction. As a control, an APH-4 Δ144-182 mutant in which the COOH end of the protein was selectively deleted without altering the bZIP domain led to an unchanged APH-4/c-Jun interaction.

In order to confirm these results, we next evaluated if the mutation of the two chosen leucine residues in the APH-3 and APH-4 double mutant affected its interacting potential with c-Jun. Hence, co-IP experiments were conducted in 293T cells cotransfected with expression vectors for APH-3L₁₆₇L₁₇₄-VV or APH-4L₁₁₃L₁₂₀-VV or with wild-type and deletion mutant APH-3 and APH-4 expression vectors combined with the c-Jun expression vector (Fig. 9). Co-IP analysis showed that the interaction between c-Jun and the APH-3L₁₆₇L₁₇₄-VV or APH-4L₁₁₃L₁₂₀-VV mutant, as well as with all tested deletion mutants of APH-3 and APH-4, was significantly reduced compared to the interaction between c-Jun and WT APH-3 and APH-4 (Fig. 9A and B). We also compared the extent of c-Jun-dependent transactivation of the collagenase promoter upon coexpression of the various deletion mutants, double point mutants, and WT versions. As shown in Fig. 9C, compared to WT APH-3 and APH-4, all deletion mutants and double point mutants demonstrated a comparable reduction in their capacity to increment the c-Jun transactivation potential. Hence, these results strongly suggest that APH-3/APH-4 and Jun family members interact to augment Jun-dependent transactivation, and that their association is dependent on atypical bZIP-like domains of APH proteins.

FIG 9.

Atypical leucine zipper region in APH-3 and APH-4 is necessary for increased Jun transactivation potential and its interaction. (A and B) Cell lysates were prepared 48 h posttransfection from 293T cells cotransfected with pMycAPH-3, pMycAPH-3Δ157-189, pMycAPH-3Δ157-167, pMycAPH-3Δ168-178, pMycAPH-3Δ179-189, and pMycAPH-3L₁₆₇L₁₇₄-VV (A) or pMycAPH-4, pMycAPH-4Δ103-135, pMycAPH-4Δ103-113, pMycAPH-4Δ114-124, pMycAPH-4Δ125-135, pMycAPH-4L₁₁₃L₁₂₀-VV (or the control pcDNA3.1), and pcDNAc-Jun (B). Cell lysates were immunoprecipitated using the anti-myc antibody and further compared to total cellular extracts by Western blotting with anti-c-Jun and anti-myc antibodies. (C) pMycAPH-3, pMycAPH-3Δ157-189, pMycAPH-3Δ157-167, pMycAPH-3Δ168-178, pMycAPH-3Δ179-189, pMycAPH-3L₁₆₇L₁₇₄-VV, pMycAPH-4, pMycAPH-4Δ103-135, pMycAPH-4Δ103-113, pMycAPH-4Δ114-124, pMycAPH-4Δ125-135, pMycAPH-4L₁₁₃L₁₂₀-VV, or pcDNA3.1 was cotransfected in 293T cells along with the expression vector for c-Jun, the collagenase-luciferase construct, and pRcActin-lacZ. Forty-eight h posttransfection, cells were lysed and luciferase activity was measured. Luciferase activity is presented as fold induction ± SEM from six transfected samples in cells transfected with the luciferase reporter construct, and the value for empty expression vectors was set at 1. (***, P ≤ 0.001).

DISCUSSION

While HTLV-3 and HTLV-4 are newly discovered deltaretroviruses and have a genomic structure that is similar to that of HTLV-1 and HTLV-2, little is known about their biology, pathogenic potential, and person-to-person transmissibility. So far, no diseases have been reported for the limited number of persons infected with these two viruses, and their geographic distribution has been restricted to Cameroon. In vitro characterization of these viruses may facilitate a better understanding of their biology and in vivo replication potential, which can affect their pathogenicity and transmissibility. We have recently shown that both HTLV-3 and HTLV-4 produce antisense spliced transcripts encoding new viral proteins that we have termed antisense protein of HTLV-3 (APH-3) and HTLV-4 (APH-4) (51). We also previously showed that APH-3 and APH-4 were capable of inhibiting the transactivation potential of Tax, similar to HBZ and APH-2, despite the absence of a typical bZIP domain. Since HBZ regulates the AP-1 pathway through a bZIP-dependent interaction with Jun family members (34, 35, 37), a comparative study between HBZ, APH-3, and APH-4 in terms of their modulation of the transactivation potential of Jun factors was conducted. In this report, we demonstrate that APH-3 and APH-4 activate Jun-dependent transcription through the AP-1-binding site present in the collagenase promoter by physically interacting with JunB, c-Jun, and JunD. The transactivation potential of the three Jun family members was observed in different cell lines in the context of chromatin and a minimal promoter and was mediated by the atypical bZIP-like region of APH-3 and APH-4.

AP-1-binding sites are found in multiple cellular promoters; thus, APH-3 and APH-4 might activate several cellular genes in infected cells. Other binding sites might also be targeted by Jun/APH heterodimers. In fact, we observed that APH-3 and APH-4, when coexpressed with Jun factors, activated the hTERT proximal minimal promoter, which harbors only Sp1- and c-myc-binding sites. A previous study similarly demonstrated that HBZ, in association with JunD, positively modulated hTERT promoter activity via a ternary complex involving Sp1 (48). However, JunB and c-Jun had no similar impact when HBZ was present. Hence, our data suggest that APH-3 and APH-4 allow the formation of a ternary complex with Sp1 and all of the different Jun family members. In this regard, the cellular localization of these heterodimers might dictate whether they are capable of inducing expression of the hTERT or any targeted genes, although preliminary data do not support a major relocalization of Jun factors by APH-3 or APH-4 upon coexpression (data not shown). Nonetheless, it will be necessary to investigate whether APH-3 and APH-4 do activate hTERT promoter activity via its Sp1-binding site through a similar HBZ/JunD/Sp1 complex.

When all four antisense-encoded proteins were compared, APH-4 generally showed the highest degree of transcription activation and displayed a level comparable to that of the c-Fos-dependent activation potential of c-Jun. The c-Fos/c-Jun heterodimer has much greater activation potential than the c-Jun homodimer due to dimer stabilization (40, 62), suggesting that the APH-4/c-Jun heterodimer forms a complex of higher stability than APH-3/c-Jun. Indeed, our IP experiments demonstrated higher levels of Jun factors complexed to APH-4, although protein levels of APH-3 and APH-4 in cellular extracts did not allow us to clearly come to the conclusion of such a higher interaction. More experiments are needed to confirm these interactions between APH-3/APH-4 and Jun factors and to determine if APH-4 indeed forms a more stable complex with Jun factors.

JunD transcriptional activation induced by HBZ has been reported to depend on the amino-terminal region of HBZ, which contains an AD (27, 46). Through deletion mutants of the N terminus and using a GAL4-based assay, our results suggest that the terminal amino regions do not possess an AD in both APH-3 and APH-4 and that the N terminus is not required for optimal Jun transactivation potential. Using an activation domain-deficient c-Jun expression vector, transfection experiments supported the conclusion that APH-3- and APH-4-mediated transcription activation greatly depended on the Jun activation domain. Several possibilities might account for this intriguing capacity of APH-3 and APH-4 to act positively on the transactivation potential of both Jun transcription factors. First, a modulatory domain in HBZ (EQERRE), which lies adjacent to the DNA binding domain, has been shown to be responsible in part for its JunD-activating potential (54). Such a charged amino acid-rich region also exists next to the leucine zipper-like motif of APH-3 and APH-4 (51), and the presence of a similar domain in APH-3 and APH-4 suggests an essential role of this domain for their effect on Jun factors. In fact, the APH-4 Δ1-80 mutant showed a more important decrease in its capacity to positively modulate the transactivation potential of c-Jun than the APH-4 Δ1-36 mutant, which could result from deletion of this charged amino acid region. It is also possible that APH-3 and APH-4, either directly or through the interaction with a ternary factor, increase the transactivation potential or binding of Jun to specific promoter regions. As an alternative mechanism, LXXLL-like motifs, which were previously shown to mediate HBZ interaction with CBP/p300 in an inhibitory manner (53), are present in the amino acid sequence of both APH-3 and APH-4 (51) and could instead potentiate the transactivation potential of Jun factors upon dimerization. We are currently exploring these possibilities.

We also identified the APH-3 and APH-4 domains responsible for their interaction with Jun factors. Previous in silico analyses of their amino acid sequence did not predict a classical bZIP domain (51, 63, 64). The HBZ bZIP domain is responsible for HBZ-Jun interactions and subsequent modulation of Jun-dependent transcription and further impacts viral infectivity and persistence (31, 34, 35, 65). Using deletion and site-specific mutagenesis, our results strongly suggest that the noncanonical bZIP found in APH-3 and APH-4 is functional and enables these proteins to interact and potentiate the transcriptional activation properties of all tested Jun family members. This is reminiscent of the atypical bZIP domain found in the EBV Zta transcription factor produced by the Epstein-Barr virus, which has been previously demonstrated to form a weak coiled-coil structure (66). Former reports have further indicated that the presence of less than five leucines in c-Jun and c-Fos leucine zippers is sufficient for dimer formation, and that hydrophobic or charged residues between the heptad of leucines importantly contribute to the stabilization and dimerization specificity (67). Hence, a minimum number of leucine heptads could be sufficient to maintain a functional bZIP enabling APH-3 and APH-4 to interact with other bZIP proteins. However, as APH-3 and APH-4 do not induce relocalization of c-Jun or other Jun family members (data not shown), these interactions might be weak, although they may retain an important regulatory role.

It is interesting that, in contrast to APH-2, which also has been shown to positively modulate the transactivation potential of Jun factors (68), a complex between APH-3/APH-4 and JunD was detected leading to activation of JunD-dependent transcription, similar to HBZ (46, 68). Similar differences might exist in the interaction of the various HTLV antisense proteins with other transcription factors. Indeed, a recent study has demonstrated that, although both HBZ and APH-2 were able to downregulate Tax-mediated LTR activation, this regulation was dependent on two different mechanisms (10). An LXXLL motif in the C-terminal region of APH-2 is responsible for the interaction with CREB and the resulting downregulation of LTR activation by Tax, while HBZ necessitates the presence of its bZIP domain for this interaction concomitantly with the LXXLL motif-dependent interaction with CBP/p300 (10, 52, 53). We have shown that APH-3 and APH-4, when deleted of their LZ-like domain, can inhibit Tax-driven LTR activation as potently as the WT, indicating that APH-3 and APH-4 also possess a CREB-interacting domain independent of the bZIP-like region (data not shown). As the C-terminal region of APH-2 also has been shown to interact with Tax2B, thereby inhibiting Tax2-dependent AP-1 activation (68), it will be important to determine if such an interaction is occurring for other HTLV-encoded antisense proteins with their respective Tax proteins.

Using the rabbit model, Yin et al. have demonstrated that unlike HBZ, APH-2 is dispensable for viral persistence and viral replication (10). Furthermore, APH-2 does not induce interleukin-2 (IL-2)-independent growth in an IL-2-dependent T-cell line, as opposed to HBZ (8), although both antisense genes share similar expression patterns (7, 69). Therefore, it is interesting that APH-2, like HBZ, is expressed in most HTLV-2-infected cells, and that this virus has been associated only with lymphocytosis and not with any form of leukemia (50, 70). Whether antisense proteins are important for these different clinical outcomes in infected patients is an important issue. A limited number of individuals have been found to be infected by HTLV-3 or HTLV-4, and diseases have not yet been reported in these persons (14, 15). Nonetheless, on the basis of our new results, it will be interesting to further investigate whether APH-3 or APH-4 could induce cell transformation or IL-2-independent T-cell growth. Preliminary data have shown that these proteins do not induce apoptosis, as assessed by the annexin V marker (data not shown).

In conclusion, in this study, we have demonstrated that APH-3 and APH-4 significantly increase the transactivation potential of the various Jun family members in the context of various promoters. Further, we have shown that this modulatory function is dependent on the bZIP-like domain of APH-3 and APH-4 mediating their interaction with different Jun factors. Although the exact mechanism by which these antisense proteins modulate Jun-dependent transcription needs to be investigated further, such functional comparisons between the four HTLV-encoded antisense proteins are important and likely will allow a better understanding of the unique association between HTLV-1 HBZ and ATL.