Upstream stimulatory factors binding to an E box motif in the R region of the bovine leukemia virus long terminal repeat stimulates viral gene expression.

The bovine leukemia virus (BLV) promoter is located in its 5'-long terminal repeat and is composed of the U3, R, and U5 regions. BLV transcription is regulated by cis-acting elements located in the U3 region, including three 21-bp enhancers required for transactivation of the BLV promoter by the virus-encoded transactivator Tax(BLV). In addition to the U3 cis-acting elements, both the R and U5 regions contain stimulatory sequences. To date, no transcription factor-binding site has been identified in the R region. Here sequence analysis of this region revealed the presence of a potential E box motif (5'-CACGTG-3'). By competition and supershift gel shift assays, we demonstrated that the basic helix-loop-helix transcription factors USF1 and USF2 specifically interacted with this R region E box motif. Mutations abolishing upstream stimulatory factor (USF) binding caused a reproducible decrease in basal or Tax-activated BLV promoter-driven gene expression in transient transfection assays of B-lymphoid cell lines. Cotransfection experiments showed that the USF1 and USF2a transactivators were able to act through the BLV R region E box. Taken together, these results physically and functionally characterize a USF-binding site in the R region of BLV. This E box motif located downstream of the transcription start site constitutes a new positive regulatory element involved in the transcriptional activity of the BLV promoter and could play an important role in virus replication.


The bovine leukemia virus (BLV) promoter is located in its 5-long terminal repeat and is composed of the U3, R, and U5 regions. BLV transcription is regulated by cis-acting elements located in the U3 region, including three 21-bp enhancers required for transactivation of the BLV promoter by the virus-encoded transactivator Tax BLV . In addition to the U3 cis-acting elements, both the R and U5 regions contain stimulatory sequences. To date, no transcription factor-binding site has been identified in the R region. Here sequence analysis of this region revealed the presence of a potential E box motif (5-CACGTG-3). By competition and supershift gel shift assays, we demonstrated that the basic helix-loop-helix transcription factors USF1 and USF2 specifically interacted with this R region E box motif. Mutations abolishing upstream stimulatory factor (USF) binding caused a reproducible decrease in basal or Tax-activated BLV promoter-driven gene expression in transient transfection assays of B-lymphoid cell lines. Cotransfection ex-
periments showed that the USF1 and USF2a transactivators were able to act through the BLV R region E box. Taken together, these results physically and functionally characterize a USF-binding site in the R region of BLV. This E box motif located downstream of the transcription start site constitutes a new positive regulatory element involved in the transcriptional activity of the BLV promoter and could play an important role in virus replication.
Bovine leukemia virus (BLV) 1 is a B-lymphotropic oncogenic retrovirus that infects cattle and is associated with enzootic bovine leukosis, a neoplastic proliferation of B-cells (1)(2)(3)(4)(5)(6). BLV is closely related to human T-lymphotropic viruses HTLV-I and -II. The majority of BLV-infected cattle are asymptomatic carriers of the virus. Only about 30% of BLV-infected animals develop a preneoplastic condition termed persistent lymphocytosis, with 2-5% developing B-cell leukemia and/or lymphoma after a long latency period. The virus can be experimentally transmitted to sheep, in which it causes similar pathologies, providing a helpful model to understand BLV and HTLV-induced leukemogenesis. BLV infection is characterized by viral latency in the large majority of infected cells and by the absence of viremia. These features are thought to be due to the transcriptional repression of viral expression in vivo (7)(8)(9). The latency is likely to be a viral strategy to escape the host immune response and allow tumor development.
BLV transcription initiates at the unique promoter located in the 5Ј-long terminal repeat (5Ј-LTR) of the BLV genome. The 5Ј-LTR is composed of the U3, R, and U5 regions and transcription initiates at the U3-R junction. The U3 region contains the main sites that regulate viral transcription ( Fig. 1) as follows: the promoter CAAT and TATA boxes (10,11), a glucocorticoidresponsive element (12)(13)(14)(15), and a large segment protected in DNase footprinting assays containing NF-B-related sites (16,17). Among the most important sites are three copies of an imperfectly conserved 21-bp sequence harboring in the middle a common 8-bp core sequence known as the cAMP-responsive element (CRE). At least three proteins, CRE-binding protein (CREB) and activating transcription factors-1 and -2 (ATF-1 and ATF-2), bind to these 21-bp enhancers in bovine B-lymphocytes, and the amount of generated complex correlates with the level of viral expression (18,19). The 21-bp enhancers are also called Tax-responsive elements (TxREs) because transactiva-tion of the BLV LTR by the virus-encoded transcriptional activator Tax BLV requires these enhancers. Because there is no evidence for direct binding of Tax BLV to DNA (20), it has been proposed that Tax BLV activation of transcription could be mediated, as reported for the HTLV system, through increased binding of the cellular proteins CREB, ATF-1, and ATF-2 (and possibly other factors yet to be identified) to the TxREs. Each of the 21-bp enhancers also contains an E box-homologous motif overlapping the CRE (Fig. 1). From 5Ј to 3Ј, these E boxes are referred to as E-box1, E-box2 and E-box3, respectively (5,21).
In addition to these cis-acting elements all situated in U3, we and others (10,22,23) have previously demonstrated that both the R and U5 regions of the BLV LTR, located downstream from the transcription start site, contain additional regulatory sequences stimulating BLV gene expression. An interferon regulatory factor (IRF)-binding site in U5 interacts with IRF-1 and IRF-2 and stimulates viral gene expression in the absence of Tax BLV (22). The presence of a 64-bp downstream activator sequence (DAS) at the 3Ј end of the R region has been reported (10, 23) (Fig. 1). To date however, no transcription factorbinding site has been identified in the R region.
In this report, the sequence analysis of the BLV R region revealed the presence of a potential E box site (hereafter referred to as E-box4), 5Ј-CACGTG-3Ј (nt ϩ173 to ϩ178), that conforms to the core hexanucleotide consensus sequence of an E box motif, CANNTG. E box sites bind proteins that belong to the basic helix-loop-helix (bHLH) family of transcription factors, including c-Myc, Max, USF, or TFE3 (reviewed in Refs. 24 -26). bHLH proteins contain a basic (b) DNA binding domain and a helix-loop-helix (HLH) dimerization domain; the latter allows these proteins to interact and to form homoand/or heterodimers. In the case of bHLH-ZIP-related proteins, the bHLH domain is contiguous with a second dimerization domain, a leucine zipper (ZIP), characterized by heptad repeats of leucines that occur immediately C-terminal to the bHLH motif. Precise DNA-binding site selection by individual bHLH family members is determined both by the central dinucleotide contained in the core hexamer sequence and by the flanking nucleotides (27)(28)(29)(30)(31)(32)(33)(34)(35)(36). Some of these transcription factors can act as either transactivators or repressors of gene expression, depending on the gene promoter or on their dimerization partner (33,(37)(38)(39)(40).
The objective of the present study was to characterize physically and functionally the E-box4 motif located in the BLV R region. We performed competition and supershift gel shift assays with nuclear extracts prepared either from peripheral blood mononuclear cells (PBMCs) derived from a BLV-infected sheep or nuclear extracts from the human B-lymphoid cell line Raji. We demonstrated that the bHLH transcription factors USF1 and USF2 specifically interacted with the E-box4. A 2-bp mutation (central CG to TA) and another 2-bp mutation (3Ј TG to GA) abrogated USF binding. To assess the transcriptional regulatory function of the E-box4, we tested the effect of these 2-bp mutations by transient transfection of B-lymphoid cell lines in the context of an LTR-luciferase reporter construct in presence or absence of a Tax BLV expression vector. Both mutations caused a reproducible 25% decrease in LTR-driven basal gene expression, indicating a positive functional role of the E-box4 motif in R. Ectopically expressed USF1 and USF2a proteins had an E-box4-dependent stimulatory effect on both the homologous BLV promoter and a heterologous thymidine kinase (TK) promoter containing multiple upstream E-box4 motifs. Mutation in the E-box4 impaired the responsiveness of the BLV promoter to Tax BLV but not to other activators known to up-regulate BLV expression (overexpression of CREB2, calcium/calmodulin-dependent protein kinase IV (CaMKIV), or CREB2/CaMKIV and treatment with PMA/ionomycin). Moreover, mutation in the E-box4 in combination with a mutation in the IRF site in U5 decreased the LTR basal activity more than 2-fold. The identification of the E-box4 motif represents the first transcription factor-binding site reported in the R region of BLV.

MATERIALS AND METHODS
PBMC Isolation-The animal used in this study was a BLV-seropositive adult sheep (M298) affected with persistent lymphocytosis, presenting a persistently elevated lymphocyte count and an inverted B/Tlymphocyte ratio. This sheep was housed at the Veterinary and Agrochemical Research Center (Uccle, Belgium). Blood samples were FIG. 1. Transcription factor-binding sites in the 5-LTR of the BLV genome. The 5Ј-LTR contains the "CAAT" (nt Ϫ97/Ϫ92) and "TATA" (nt Ϫ43/Ϫ37) boxes upstream of the transcription initiation site at the U3-R junction (mRNA start site at ϩ1 is indicated by an arrow). The TxREs are three imperfectly repeated sequences of 21 bp. These are major transcriptional enhancers, which interact with the cellular transcription factors CREB, ATF-1, and ATF-2. Transcriptional activation of the BLV LTR by the viral encoded Tax BLV transactivator requires these enhancers. Moreover, the activity of the factors CREB, ATF-1, and ATF-2 is increased by protein kinases A and CaMKIV. Each of the 21-bp enhancers contains a sequence homologous to the consensus E box-binding motif (referred to as E-box1, E-box2, and E-box3) overlapping an imperfect CRE (referred to as CRE1, CRE2, and CRE3). The U3 region also contains a glucocorticoid-responsive element (GRE) and a large segment protected in DNase footprinting assays (nt Ϫ118/Ϫ70) containing NF-B-related sites. Outside of U3, a DAS (nt ϩ147/ϩ210) in the R region and an IRF-binding site (nt ϩ246/ϩ269) in the U5 region, interacting with IRF-1 and IRF-2, are regulatory sequences important for BLV gene expression. The E box motif identified in the present report and located in the R region is also indicated and is referred to as E-box4.
collected by jugular venipuncture, mixed with EDTA as an anticoagulant, and centrifuged at 1750 ϫ g for 25 min at room temperature. The PBMCs were then isolated by Percoll gradient centrifugation (density 1.129 g/ml; Amersham Biosciences) and washed twice in phosphatebuffered saline containing 0.075% EDTA, with centrifugation steps at 450 ϫ g for 10 min at room temperature. The cells were washed with phosphate-buffered saline (centrifugations at 200 ϫ g for 10 min at room temperature) until the supernatant became clear.
Cell Lines and Cell Culture-The Raji and Daudi cell lines are human B-lymphoid EBV-positive cell lines derived from Burkitt's lymphomas. The human epithelial HeLa cell line is derived from a cervical carcinoma and is transformed by human papilloma virus type 18. All media, sera (Myoclone Superplus), and supplements were from Invitrogen. Raji cells were grown in RPMI 1640-Glutamax I medium supplemented with 10% fetal bovine serum, 50 units of penicillin/ml, and 50 g of streptomycin/ml. Daudi cells were maintained in RPMI 1640-Glutamax I medium with 10% fetal bovine serum, 10 mM HEPES buffer, 1 mM sodium pyruvate, 50 units of penicillin/ml, and 50 g of streptomycin/ml. HeLa cells were cultured in Dulbecco's modified Eagle's-Glutamax I medium containing 5% fetal bovine serum, 50 units of penicillin/ml, and 50 g of streptomycin/ml. All cells were grown at 37°C in an atmosphere of 5% CO 2 .
The p(E-box4) 3 TK-luc and p(E-box4) 7 TK-luc were generated by inserting three direct repeats in the forward orientation and seven direct repeats in the reverse orientation, respectively, of an oligonucleotide with the sequence 5Ј-ACCGTCTCCACGTGGACTCTCT-3Ј (the BLV E-box4 motif is underlined) into SmaI-digested pTK-luc. Similarly, the two mutated versions, p(E-box4-mutA) 3 TK-luc and p(E-box4-mutB) 3 TK-luc, were generated by inserting three direct repeats in the forward orientation of an oligonucleotide with the sequence 5Ј-ACCG-TCTCCATATGGACTCTCT-3Ј and 5Ј-ACCGTCTCCACGGAGACTCT-CT-3Ј (mutations are highlighted in bold), respectively, into SmaIdigested pTK-luc. In all these constructs, the multimerized wild-type or mutated E box motifs are thus positioned upstream of the TK promoter.
To construct pGEM-LTR BLV used in RNase protection analysis, a 201-bp fragment containing the BLV 5Ј-LTR from position Ϫ118 to ϩ83 was generated by PCR amplification of pLTRwt-luc. XbaI and PstI restriction sites were introduced into the 5Ј and 3Ј PCR primers, respectively, and the XbaI-PstI-restricted PCR fragment was cloned in pGEM-3Zf(ϩ) vector (Promega) digested with XbaI and PstI. The 5Ј primer oligonucleotide corresponding to nt Ϫ118 to Ϫ87 contained an added XbaI site (in bold) at the 5Ј end (5Ј-CGCTCTAGAG nt-118 GCTA-GAATCCCCGTACCTCCCCAACTTCCCC-3Ј). The 3Ј primer oligonucleotide corresponding to nt ϩ55 to ϩ83 contained an added PstI site (in bold) at the 5Ј end (5Ј-GGTTTTCTGCAGC ntϩ83 GGATAGCCGACCAG-AAGGTCTCGGGAGC-3Ј).
To construct the expression plasmid for ovine c-Myc, we used PCR to amplify the coding region of the ovine c-myc cDNA described in Kiermer et al. (43). HindIII and BamHI restriction sites were introduced into the 5Ј and 3Ј PCR primers, respectively, and the HindIII-BamHI-restricted PCR fragment was cloned in pcDNA3.1 (ϩ/Ϫ) vector (Invitrogen) digested with HindIII and BamHI. The 5Ј primer oligonucleotide corresponding to nt 203-226 (according to the ovine c-Myc cDNA sequence, GenBank TM accession number Z68501) contained an added HindIII site (in bold) at the 5Ј end (5Ј-GCCCAAGCTTA nt203 CCGCGATGCCCCTC-AACGTCAGC-3Ј). The 3Ј primer oligonucleotide corresponding to nt 1571-1598 contained an added BamHI site (in bold) at the 5Ј end (5Ј-GCCGGATCCT nt1598 CCTCACTTTCCCTTAGTAACAAATGAG-3Ј). The resulting plasmid was designated pcDNA3-cMyc.
The eukaryotic expression vectors pSG-TAX BLV and pSG-CREB2 (44) were gifts from Drs. Luc Willems and Richard Kettmann. The expression plasmid pCaMKIV was a gift from Dr. Anthony Means (45).
The USF1 and USF2a expression vectors (kindly provided by Drs. A. Kahn and M. Raymondjean) contained the human USF1 and USF2a cDNAs cloned in the pCR3 expression vector parent plasmid (Invitrogen) and were described previously (46).
Transient Transfection and Luciferase Assays-Raji cells were transfected by using the DEAE-dextran procedure as described previously (47). Briefly, cells were harvested at density of 10 6 /ml, washed with STBS (25 mM Tris-HCl (pH 7.5), 137 mM NaCl, 5 mM KCl, 700 M CaCl 2 , 500 M MgCl 2 , 600 M Na 2 PO 4 ), and resuspended at a concentration of 6 ϫ 10 6 in 600 l of a mixture containing 500 ng of the pGL3-BASICderived constructs (with or without cotransfected DNAs) and 450 g of DEAE-dextran (Amersham Biosciences)/ml in STBS. Cells were incubated for 1 h at 37°C, washed twice with STBS and once with culture medium, and grown in 3 ml of supplemented medium for 40 -44 h. Cells were then lysed and assayed for luciferase activity (Promega). Luciferase activities derived from BLV LTRs were normalized with respect to protein concentrations using the Detergent-compatible Protein Assay (Bio-Rad).
Daudi cells were transfected by electroporation. Cells were harvested in exponential growth phase and resuspended in supplemented medium at a concentration of 5 ϫ 10 6 per 400 l. The 400 l of cells were mixed with 8 g of pGL3-BASIC-derived constructs and 50 ng of pRL-TK (Promega) and incubated for 15 min at room temperature, transferred to electroporation vials, and electroporated at 250 V with a capacitance of 960 microfarads (by using a Bio-Rad gene pulser). Transfected cells were collected, plated out immediately in 4 ml of preheated medium (a 1:1 mixture of fresh culture medium and supernatant of a 24-h culture), and grown for 48 h at 37°C. All transfection mixtures contained the pRL-TK, in which a cDNA encoding Renilla luciferase is under the control of the herpes simplex virus thymidine kinase promoter region and is used as an internal control for transfection efficiency. At 48 h post-transfection, luciferase activities (firefly and Renilla) were measured in cell lysates, and firefly luciferase activities derived from the BLV promoters were normalized with respect to the Renilla luciferase activity by using the dual luciferase reporter assay system (Promega).
HeLa cells were transfected using FuGENE TM -6 (Roche Molecular Biochemicals) according to the manufacturer's protocol. Briefly, cells were seeded at a density of 2 ϫ 10 5 cells/well in 6-well plates. Fu-GENE TM -6 was added directly to serum-free medium 5 min prior to addition to the DNA. Hundred microliters of this FuGENE TM -6/serumfree medium mix was added to the DNA mixture in microcentrifuge tubes. The mixture was incubated for 15 min at room temperature and, finally, was added to each well of a 6-well plate. Transfected cells were grown in 2 ml of supplemented medium for 40 -44 h. Cells were then lysed and assayed for luciferase activity (Promega). Luciferase activities derived from BLV or TK promoters were normalized with respect to protein concentrations using the Detergent-Compatible Protein Assay (Bio-Rad).
Electrophoretic Mobility Shift Assays (EMSAs)-Nuclear extracts were prepared by a rapid method described by Osborn et al. (48). All buffers contained the protease inhibitors antipain (10 g/ml), aprotinin (2 g/ml), chymostatin (10 g/ml), leupeptin (1 g/ml), and pepstatin (1 g/ml). Protein concentrations were determined by the method of Bradford (49). The DNA sequences of the coding strand of the doublestranded oligonucleotides used for this study are listed in Fig. 3A or in the figure legends. EMSAs were performed as described previously (47). Briefly, nuclear extract (4 g of protein) was first incubated at room temperature for 10 min in the absence of probe and specific competitor DNA in a 16-l reaction mixture containing 10 g of DNase-free bovine serum albumin (Amersham Biosciences), 6 g of poly(dI-dC) (Amersham Biosciences) as nonspecific competitor DNA, 1 mM dithiothreitol, 20 mM Tris-HCl (pH 7.5), 60 mM KCl, 1 mM MgCl 2 , 0.1 mM EDTA, and 10% (v/v) glycerol. Fifteen thousand cpm of probe (10 -40 fmol) was then added to the mixture with or without a molar excess of an unlabeled specific DNA competitor, and the mixture was incubated for 20 min at room temperature. Samples were subjected to electrophoresis at room temperature on 6% polyacrylamide gels at 150 V for 2-3 h in 1ϫ TGE buffer (25 mM Tris acetate (pH 8.3), 190 mM glycine, 1 mM EDTA). Gels were dried and autoradiographed for 24 -48 h at Ϫ70°C. For supershift assays, polyclonal antibodies against USF1 (sc-229), USF2 (sc-862), Max (sc-197), Mad-1 (sc-222), Mad-2 (sc-1720), Mad-3 (sc-933), Mad-4 (sc-771), c-Myc (sc-764), and Mnt (sc-769) (Santa Cruz Biochemicals, Santa Cruz, CA) were added at a final concentration of 2 g/ reaction to the binding reaction mixture at the end of the binding reaction for an additional 30 min incubation at room temperature before electrophoresis.
In vitro translated Ebox-binding proteins Max, Mad1, and c-Myc were used in control EMSAs. These proteins were generated by using the TNT-coupled Reticulocyte Lysate System (Promega) with the pGEM expression plasmid encoding mouse Max (p22 long form), the pRC/CMV expression vector for human Mad1 (both kindly provided by Dr. S. Segal), and the pcDNA3 expression vector for ovine c-Myc, pcDNA3-cMyc. Eight l of in vitro translated Mad1 or 4 l of in vitro translated c-Myc were preincubated with 1 l of in vitro translated Max for 10 min at 42°C and then for an additional 20-min period at room temperature to promote formation of Mad1-Max or c-Myc-Max dimers, respectively. These complexes were then incubated for 30 min at room temperature in the presence of 30,000 cpm of a 32 P-labeled CMD (specific E box) probe (an oligonucleotide with a c-Myc consensus binding site, 5Ј-AGCTTCAGACCACGTGGTCGGG-3Ј (50)) in a buffer containing 10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 0.5 mM dithiothreitol, 0.05% Nonidet P-40, 1% glycerol, and 1 g of poly(dI-dC) in a total volume of 20 l. For supershift assays, 2 l of either anti-Max (sc-197), either anti-Mad-1 (sc-222), or anti-c-Myc (sc-764) antibodies (Santa Cruz Biochemicals, Santa Cruz, CA) were added to the reaction mixture and further incubated overnight at 4°C. The protein-DNA complexes were then separated from the free probe by electrophoresis on 4% polyacrylamide gels in 0.5ϫ Tris borate/EDTA at 150 V. The free probe was run out of the gel for better separation of the complexes.
RNase Protection Assays-Raji cells were harvested by centrifugation 44 h post-transfection, washed in phosphate-buffered saline, pelleted, and kept on ice. Total RNA samples were prepared using the commercial RNAqueous Phenol-free Total RNA Isolation Kit (Ambion) and treated with 2 units/g of RNase-free DNase I (10 units/l, Roche Molecular Biochemicals) for 30 min at 37°C in a total volume of 240 l containing 1ϫ NEB2 buffer (New England Biolabs) and 1.6 units/g RNasin (40 units/l, Promega). RNA was then extracted with phenol/ chloroform/isoamyl alcohol, ethanol-precipitated, and resuspended in sterile and RNase-free water (United States Biochemical Corp.).
A BLV-specific 32 P-labeled antisense riboprobe was synthesized in vitro by transcription of XbaI-restricted pGEM-LTR BLV with SP6 polymerase according to the protocol provided with the Riboprobe in vitro Transcription Systems (Promega). A luciferase antisense riboprobe was similarly synthesized by transcription of SgrAI-restricted pSP-lucϩ vector (Promega) with T7 polymerase. As control, a glyceraldehyde-3phosphate dehydrogenase (GAPDH)-specific antisense probe was synthesized by the same method and used on the same RNA samples.
The RNase protection assays were performed with the RPA II Kit (Ambion) according to the manufacturer's recommendations. Briefly, hybridization reactions (20 l) containing 120 g of total cellular RNA and 200,000 cpm of probe in hybridization buffer were heated to 95°C for 4 min to denature the RNA and then incubated at 42°C for 16 h. The reaction mixtures were diluted by the addition of 200 l of digestion buffer, and the single-stranded sequences were digested with RNase T1 and RNase A for 1 h at 37°C. Following addition of 300 l of inactivation buffer and ethanol precipitation, the protected RNA fragments were analyzed by electrophoresis through 6% urea polyacrylamide gels.

RESULTS
The BLV R Region E-box4 Motif Specifically Binds USF1 and USF2 in Vitro-A data base search for potential transcription factor-binding sites in the R region of the BLV LTR revealed the presence of a putative E box site (nt ϩ173/ϩ178 according to the sequence reported by Sagata et al. (41)), 5Ј-CACGTG-3Ј. In order to identify cellular factors that bind to this potential E box motif (designated E-box4), EMSAs were performed by using as probe a 22-bp E-box4-containing oligonucleotide corresponding to positions ϩ165 to ϩ186 of the 5Ј-LTR. This probe (referred to as E-box4-wt) was incubated with nuclear extracts from PBMCs derived from a BLV-infected sheep (BLV-infected sheep M298 presenting a persistently elevated lymphocyte count and an inverted B/T-lymphocyte ratio). A major retarded protein-DNA complex was detected with a major band indicated by an arrow and the fainter band indicated by an asterisk (Fig. 2A, lane 1). To evaluate the specificity of these interactions, unlabeled double-stranded oligonucleotides were prepared and used as competitors in the EMSAs. Binding of proteins in the major complex was shown to be sequence-specific by competition EMSAs. Indeed, the formation of this complex was competed for by an excess of the unlabeled homologous E-box4-wt oligonucleotide ( Fig. 2A, lanes 2-5) but not by the same molar excess of a heterologous oligonucleotide of unrelated sequence containing the HIV-1 NF-B sites ( Fig. 2A,  lanes 10 -13). A well characterized E box motif from the HIV-1 promoter (51,52) was also used as a competitor and inhibited the formation of the major complex as efficiently as the homologous oligonucleotide ( Fig. 2A, lanes 6 -9), demonstrating that the major complex was specific to the E box motif. In contrast, the specificity of the fainter and slower migrating band was uncertain because this band was competed by all oligonucleotides tested (the homologous and the heterologous oligonucleotides, Fig. 2A, and the two mutated E-box4 oligonucleotides, see below Fig. 3B), except by the HIV-1 E box oligonucleotide ( Fig. 2A, lanes 6 -9). Moreover, this fainter band was not affected by any of the E box-specific antibodies we used in supershift assays (Fig. 2B, see below) and was not further characterized. Similar results were obtained with nuclear extracts from the Raji cell line, a human B-lymphoid cell line (data not shown).
To identify directly the factors present in the specific major retarded complex, we performed supershift assays using antibodies directed against individual members of the bHLH family of transcription factors (Fig. 2B, left panel). Labeled E-box4-wt oligonucleotide probe was incubated with nuclear extracts from BLV-infected ovine PBMCs. Polyclonal antibodies directed against USF1, USF2, Max, Mad-1, Mad-2, Mad-3, Mad-4, c-Myc, and Mnt were added to the binding reaction mixture. Addition of either the anti-USF1 or the anti-USF2 antibody (Fig. 2B, left panel, lanes 2 or 3, respectively) interfered with the formation of the major complex and generated supershifted complexes of decreased mobility (Fig. 2B, left panel). Although the anti-USF1 antibody eliminated almost completely the major complex, the anti-USF2 antibody only slightly decreased the intensity of this complex. When both anti-USF1 and anti-USF2 antibodies were included in the binding reaction, the entire complex was eliminated (Fig. 2B,  left panel, lane 4). In contrast, the binding pattern was not affected by the addition of the antibodies directed against the other bHLH proteins (Fig. 2B, left panel, lanes 5-11), showing that the major retarded complex did not seem to involve these other proteins. No supershifted complex was observed with a control purified rabbit IgG (Fig. 2B, left panel, lane 1) or when the antibodies were added to the probe alone (data not shown), thus indicating the specificity of the protein-antibody interactions. Similar results were observed with Raji nuclear extracts (data not shown).
In order to establish the validity and/or the specificity of some of the antibodies other than anti-USF1/2 used in the supershift assays, control EMSAs were performed by incubating in vitro translated E box-binding proteins (Max, Mad1, and c-Myc) with a labeled CMD (specific E box) probe (50) (Fig. 2B Overall, our results demonstrate that both the USF1 and USF2 transcription factors interact with the E-box4 motif located in the BLV R region. Moreover, they suggest that the  10 -13). The sequence of the coding strand of the HIV-1 E box oligonucleotide is 5Ј-ATTTCATCACGTGGCCCGAG-3Ј (nt ϩ281 to ϩ300 according to the numbering of the HIV-1 BRU genome where nt ϩ1 is the start of U3 in the 5Ј-LTR; the E box motif is underlined) (52). The HIV-1 NF-B oligonucleotide contains the two NF-B-binding sites of the HIV-1 enhancer (nt ϩ350 to ϩ373) and has been described previously (136). The major DNA-protein complex and the fainter band are indicated by an arrow and an asterisk, respectively. The free probe (FP) is also indicated. B, supershift assays. Left panel, the E-box4-wt oligonucleotide probe was incubated with 4 g of nuclear extracts from BLV-infected ovine PBMCs. Next, antibodies directed against different members of the bHLH family of transcriptional regulatory proteins (lanes 2-11) or purified rabbit IgG as a negative control (lane 1) were added to the binding reaction. An E box-binding protein constitutively present in the reticulocyte lysate is indicated by an asterisk. The supershifted complexes are also indicated. The free probe was run out of the gel for better separation of the complexes.
predominant USF species that bind to this motif are the heterodimer USF1/USF2 and the homodimer USF1/USF1 and that a small amount of homodimer USF2/USF2 might also bind. These data are consistent with several reports (53)(54)(55)(56) demonstrating that the major USF species present in most tissues and cell lines is the heterodimer between USF1 and USF2. USF1 homodimers are less abundant, and USF2 homodimers are usually quite scarce.
Identification of Point Mutations Abolishing USF Binding to the E-box4 Motif in the 5Ј-LTR BLV R Region-To further characterize physically the E-box4 motif located in the R region of the BLV 5Ј-LTR, we studied by EMSA the effect of selected mutations on binding affinity. Two mutations were designed to abolish binding of factors to the E-box4 motif. The first mutation, designated E-box4-mutA, consisted of the substitution of the two central nucleotides CG with the dinucleotide TA (57). The second mutation, designated E-box4-mutB, consisted of the substitution of the 3Ј-TG with the dinucleotide GA (58) (Fig. 3A).
The effects of these 2-bp mutations on binding affinity were analyzed by competition EMSAs with the E-box4-wt oligonucleotide as a probe and nuclear extracts from sheep PBMCs (Fig. 3B). The E box-specific retarded complex was inhibited by competition with an excess of the homologous oligonucleotide (Fig. 3B, lanes 2-5). In contrast, the complex was not competed by the E-box4-mutA and E-box4-mutB oligonucleotides (Fig.  3B, lanes 6 -9 and lanes 10 -13, respectively), demonstrating that both mutations abolished USF binding to the E-box4. Moreover, we confirmed the lack of USFs binding to probes corresponding to the mutated E-box4 motifs (data not shown). Similar results were observed with Raji nuclear extracts (data not shown). Taken together, our results identify an E box motif in the R region of the BLV 5Ј-LTR (nt ϩ173 to ϩ178). This E box motif specifically binds the bHLH proteins USF1 and USF2 in vitro. We report a couple of 2-bp mutations, referred to as E-box4-mutA and E-box4-mutB, that abrogate USF binding to this motif.
The E-box4 Motif in the R Region of the BLV 5Ј-LTR Is Required for Optimal Basal Gene Expression from the BLV Promoter-In order to examine the functional role of the E-box4 motif in the basal transcriptional activity of the BLV promoter, the same 2-bp mutations described above (E-box4-mutA and E-box4-mutB) were introduced by site-directed mutagenesis in the context of a pLTR BLV -luciferase reporter construct, called pLTRwt-luc and containing the firefly luciferase (luc) gene under the control of the complete 5Ј-LTR (nt Ϫ211 to ϩ320) of the 344 BLV provirus. Strain 344 is an infectious and pathogenic molecular clone. The two mutated plasmids were designated pLTR(E-box4-mutA)-luc and pLTR(E-box4-mutB)luc, respectively.
To assess the transcriptional regulatory function of the E-box4 motif, the constructs pLTRwt-luc, pLTR(E-box4-mutA)-luc, and pLTR(E-box4-mutB)-luc were transiently transfected into Raji cells. At 44 h post-transfection, cells were lysed and assayed for luciferase activity. Results presented in Fig. 4A show that both mutations, E-box4-mutA and E-box4-mutB, reproducibly reduced the basal activity of the BLV promoter by 25%. These results are consistent with the binding of a transcriptional activator, such as USF1 and/or USF2, to the E-box4 motif.
To confirm these results in another B-lymphoid cell line, the same reporter constructs (pLTRwt-luc, pLTR(E-box4-mutA)luc, and pLTR(E-box4-mutB)-luc) were transiently transfected into the human cell line Daudi. As shown in Fig. 4B, the E-box4-mutA mutation caused a 15% decrease in basal LTRdirected gene expression. Interestingly, the E-box4-mutB mutation more strongly decreased LTR activity because it caused a 56% decrease in BLV 5Ј-LTR-driven basal gene expression. These functional results confirmed in another B-cell line the results observed in Raji cells. However, the E-box4-mutA and E-box4-mutB mutations decreased LTR basal activity with less strength and more strength, respectively, in Daudi cells than observed in Raji cells. These differential effects of both mutations between the two cell lines may result from differences in the expression, binding, and/or transcriptional activity of USF proteins, transcriptional cofactors, and/or other DNA-binding proteins.
Our functional results thus demonstrate a positive regulatory role of the E-box4 motif located in the R region of the BLV 5Ј-LTR in Tax-independent BLV promoter-driven gene expression.
Ectopic Expression of USF1 and USF2a Transcription Factors Up-regulates BLV Promoter Activity in Part through the E-box4 Motif-To examine the role of USF in regulation of the BLV promoter, we studied the effect of overexpression of USF on luciferase activity. Cotransfection of expression constructs FIG. 3. Mutagenesis of the E-box4 motif located in the BLV R region. A, nucleotide sequence of the E-box4-wt oligonucleotide. The underlined bases indicate the E box motif, which is aligned with the E box consensus sequence. For the E-box4-mutA and E-box4-mutB oligonucleotides, only the bases that are changed compared with the wt sequence are indicated. B, E-box4-wt and mutated oligonucleotides were tested in competition EMSAs. The E-box4-wt oligonucleotide was 5Ј-end-labeled and used as probe. This probe was incubated with 4 g of nuclear extracts from BLV-infected ovine PBMCs either in the absence of competitor (lane 1) or in the presence of increasing concentrations (5-, 10-, 20-, and 80-fold molar excess) of the homologous E-box4-wt oligonucleotide (lanes 2-5), of the E-box4-mutA oligonucleotide (lanes 6 -9), or of the E-box4-mutB oligonucleotide (lanes 10 -13). The major DNA-protein complex and the fainter band are indicated by an arrow and an asterisk, respectively. The free probe (FP) is also indicated. for USF1 and/or USF2a with the pLTRwt-luc reporter construct had no significant effect on luciferase activity in the transformed Raji cell line (data not shown). Considering that USF is abundantly expressed in most cell types, one possible explanation is that USFs would be non-limiting factors for the BLV promoter activity in Raji cells. Moreover, recent studies comparing USF function in different normal and cancerous breast cell lines have demonstrated that a partial or complete loss of USF transcriptional activity is a common event in transformed cells as compared with normal cells, even though there is no difference both in expression and DNA-binding activity of USF proteins (56,59). These latter studies suggest a correlation between the loss of USF function and tumorigenesis. Therefore, the potential inactivation of USF in the transformed Raji cells we used is another possible explanation. Furthermore, USF was originally identified and characterized in HeLa cells (30, 60 -64), and it has been demonstrated that USF proteins are transcriptionally active in this cell line (59).
Together, these observations prompted us to examine in HeLa cells whether USF1 or USF2a overexpression had an effect mediated by the E-box4 on BLV promoter activity. We performed cotransfection experiments using USF1 or USF2a expression vectors and the pLTRwt-luc reporter construct as well as the mutated derivative pLTR(E-box4-mutB)-luc. As shown in Fig. 5, USF1 and USF2a transactivated the BLV promoter in a dose-dependent manner up to 4-and 3.2-fold, respectively. Maximal activation was seen at concentrations of 2000 ng of USF1 (Fig. 5A) and 250 ng of USF2a (Fig. 5B). When the amounts of USF2a were further increased, the stimulation was reduced. When expressed at very high levels, many transcriptional activators squelch activated transcription (65). This squelching phenomenon is thought to be due to the sequestration in solution of transcriptional components, preventing their interaction at gene promoters. Moreover, cotransfection of both USF1 and USF2a together in equal amounts resulted in transactivation levels of the BLV promoter similar to those obtained with USF1 and USF2a cotransfected separately (data not shown). This observation has also been reported in numerous other studies (66 -71) and is consistent with the idea that USF homodimers can function as well as heterodimers, although in purified nuclear extracts USFs were generally found as heterodimers (72).
However, when analyzing in the same cotransfection assays the effect of USF1 and USF2a overexpression on the mutated construct pLTR(E-box4-mutB)-luc, we observed that mutation of the E-box4 did not reduce the USF stimulatory effect. Indeed, USF1 and USF2a transactivated pLTR(E-box4-mutB)luc as efficiently as the wild-type construct pLTRwt-luc (Fig. 5,  A and B), indicating that other LTR region(s) could mediated partially or totally transactivation by USF. These other regions could mask the potential transactivation mediated by the E-box4 motif. Because three E boxes are located in the BLV LTR U3 region (E-box1, -2, and -3, Fig. 1), we evaluated the potential contribution of these motifs in the LTR response to USF1 and USF2a. Mutant reporter constructs were generated, in which the E-box1, -2, and -3 were mutated and in which all four E boxes were mutated. The plasmids were designated pLTR(E-box1,2,3-mutB)-luc and pLTR(E-box1,2,3,4-mutB)-luc, respectively, and were used in cotransfection experiments of HeLa cells with the USF1 and USF2a expression vectors. As shown in Fig. 5, A and B, mutation of the E boxes 1-3 in combination attenuated the stimulatory effect of USF1 and USF2a. There was a 28 (USF1) and 39% (USF2a) reduction of stimulation compared with the USF stimulations observed with pLTRwtluc. Interestingly, when all four E boxes were mutated, we observed a 53 (USF1) and 57% (USF2a) reduction of stimulation, suggesting that USF proteins act in part through the E-box4 motif to stimulate BLV promoter activity. It is noteworthy that, although a mutated LTR in which all four E boxes were mutated was strongly impaired in terms of stimulation by USF, this mutated LTR was still responsive to ectopic USF. This residual response may result either from unidentified E boxes present in the LTR and/or from USF action directly or indirectly through other cis-regulatory elements.
These results thus indicate that the E-box4 plays a role in the transactivation of the BLV promoter by USF but that the three other E boxes located in U3 are also responsible for part of the USF response. Moreover, besides these four E boxes, other cis-elements that are directly or indirectly affected by USF seem to contribute to the residual USF induction observed with the total mutant construct pLTR(E-box1,2,3,4-mutB)-luc.
Multimerized Copies of the BLV E-box4 Motif Confer USF Inducibility to a Heterologous Minimal Promoter-To assess better the contribution of the E-box4 to the ectopic USF response, we constructed three luciferase reporter plasmids driven by the herpes simplex virus TK minimal promoter with or without three or seven tandem repeats of the E-box4 motif inserted upstream of the TK promoter. These three plasmids were referred to as pTK-luc, p(E-box4) 3 TK-luc, and p(E-box4) 7 TK-luc, respectively. Moreover, three copies of the E-box4-mutA and E-box4-mutB oligonucleotide were also cloned into pTK-luc and called p(E-box4-mutA) 3 TK-luc and p(E-box4-mutB) 3 TK-luc, respectively. To examine the USF inducibility of these reporter constructs, HeLa cells were transiently cotransfected with each of them and increasing amounts of either the USF1 or USF2a expression vector (Fig. 6, A or B, respectively) and then assayed for luciferase activity. The control pTK-luc construct was moderately transactivated by USF1 (up to 3.5-fold) and by USF2a (up to 7.7-fold). Cotransfection of USF1 or USF2a expression vectors with the reporter constructs containing the wild-type E-box4 motifs resulted in a stimulation of luciferase activity that was dependent on the number of E boxes [p(E-box4) 3 TK-luc versus p(E-box4) 7 TK-luc]. Addition of three copies of the E-box4 motif upstream of the TK promoter resulted in a dose-dependent increase in luciferase activity by ectopically expressed USF1 (up to 8.2-fold) and FIG. 5. Response of the BLV promoter to ectopically expressed USF transcription factors. HeLa cells were transiently cotransfected with 500 ng of an LTR luciferase reporter construct wild type or mutated in the different E boxes (as indicated) and increasing amounts (0, 100, 250, 500, 1000, and 2000 ng) of either the USF1 or the USF2a expression vector, pCR3-USF1 or pCR3-USF2a (A or B, respectively). To maintain the same amount of transfected DNA and to avoid squelching artifacts, the different amounts of USF expression vector cotransfected were complemented to 2000 ng of DNA by using the empty pCR3 vector. Luciferase activities were measured in cell lysates 44 h after transfection and normalized to protein concentrations. Results are presented as histograms indicating the induction by USF1 or USF2a (in fold) with respect to the activity of each LTR reporter construct in the absence of USF, which was assigned a value of 1. Means and standard errors of the means are shown. A representative experiment of four independent transfections is shown.
FIG. 6. Ability of multimerized BLV E-box4 motifs to confer USF stimulation to a TK minimal promoter. HeLa cells were transiently cotransfected with 500 ng of a TK luciferase reporter construct devoid or not of upstream wild-type or mutated E-box4 motifs (as indicated) and increasing amounts (0, 100, 250, 500, 1000, and 2000 ng) of either the USF1 or the USF2a expression vector, pCR3-USF1 or pCR3-USF2a (A or B, respectively). To maintain the same amount of transfected DNA and to avoid squelching artifacts, the different amounts of USF expression vector cotransfected were complemented to 2000 ng of DNA by using the empty pCR3 vector. Luciferase activities were measured in cell lysates 44 h after transfection and normalized to protein concentrations. Results are presented as histograms indicating the induction by USF1 or USF2a (in fold) with respect to the activity of each TK reporter construct in the absence of USF, which was assigned a value of 1. Means and standard errors of the means are shown. A representative experiment of six independent transfections is shown.
USF2a (up to 21.1-fold), thus representing a 2.3-and 2.7-fold up-regulation when compared with the USF response of the control pTK-luc devoid of upstream E-box4 motifs. Seven copies of the E-box4 further increased the USF stimulation: 4.5-(USF1) and 4-fold (USF2a) up-regulation compared with the pTK-luc. This effect required intact E-box4 motifs, because mutations in these motifs [p(E-box4-mutB) 3 TK-luc (Fig. 6, A  and B) and p(E-box4-mutA) 3 TK-luc (data not shown)] resulted in levels of USF-mediated transactivation similar to those obtained with the control pTK-luc. Moreover, using each TK reporter construct, the combination of USF1 and USF2a transfected together in equal amounts had no more effect than either factor transfected alone (data not shown).
We conclude from these experiments that ectopic USF1 and USF2a proteins have an E-box4-dependent stimulatory effect on the heterologous TK promoter containing multiple upstream E-box4 motifs. These results thus establish the functional significance of USF through the BLV E-box4 motif.

The E-box4 Motif in the R Region of the BLV 5Ј-LTR Regulates Tax-dependent BLV Promoter-driven Gene Expression-
Efficient transcription and replication of the BLV genome require both the viral LTR and the virus-encoded transcriptional activator Tax BLV , which functions through the TxREs in U3. In the next experiments, we studied the effect of the E-box4-mutA and E-box4-mutB mutations on the responsiveness of the BLV promoter to Tax BLV . Toward this end, cotransfections of Raji cells were performed using either pLTRwt-luc, pLTR(E-box4-mutA)-luc, or pLTR(E-box4-mutB)-luc and increasing amounts of the Tax BLV expression vector, pSG-TAX BLV . At 44 h posttransfection, luciferase activity was measured in cell lysates. As shown in Fig. 7A, Tax BLV activation of the wild-type LTR ranged from 44.3-to 214-fold. In comparison, Tax BLV activation of the pLTR(E-box4-mutA)-luc and of the pLTR(E-box4-mutB)luc ranged from 32.6-to 160-fold and from 24.7-to 126-fold, respectively. Thus, the E-box4-mutA and E-box4-mutB mutations reproducibly reduced Tax BLV -mediated transactivation Means and standard errors of the means from six independent transfections performed with at least two different DNA preparations are shown. B, Raji cells were transiently cotransfected with 500 ng of either pLTRwt-luc, pLTR(E-box4-mutA)-luc, or pLTR(E-box4-mutB)-luc and increasing amounts of both pSG-CREB2 (0, 50, 150, 300, and 600 ng) and pCaMKIV (0, 50, 150, 300, and 600 ng). To maintain the same amount of transfected DNA and to avoid squelching artifacts, the different amounts of pSG-CREB2-pCaMKIV cotransfected were complemented to 1200 ng of DNA by using the empty plasmids pSG5 and a pCMV-based vector, respectively. Luciferase activity was measured in cell lysates 44 h after transfection. The results are presented as histograms indicating luciferase activities (arbitrary units) normalized to protein concentrations (top panel) or indicating the induction by CREB2/CaMKIV (in fold) with respect to the activity of the same reporter construct in the absence of cotransfected expression vectors, which was assigned a value of 1 (bottom panel). Means and standard errors of the means from four independent transfections performed with at least two different DNA preparations are shown. by 21-26 and by 29 -44%, respectively. These results demonstrate a positive regulatory role of the E-box4 motif in Tax-dependent BLV promoter-driven gene expression.
We next wanted to test the effect of the E-box4-mutA and E-box4-mutB mutations on the activation of the BLV LTR by other stimuli. It has been reported previously (19,44) by transient transfections of D17 osteosarcoma cells that the bovine CREB2 protein is able to transactivate the BLV LTR in the absence of the Tax BLV protein and that the cAMP-dependent protein kinase A or the calcium/calmodulin-dependent protein kinase IV (CaMKIV) substantially increases the ability of CREB2 to stimulate gene expression. Moreover, ex vivo, BLV expression can be up-regulated by several lymphocyte activators, including phorbol ester PMA (73) and calcium ionophore ionomycin (74).
We performed transient cotransfections of Raji cells with either pLTRwt-luc, pLTR(E-box4-mutA)-luc, or pLTR(E-box4-mutB)-luc and increasing amounts of both the CREB2 expression vector, pSG-CREB2, and the CaMKIV expression vector pCaMKIV. At 44 h post-transfection, luciferase activity was measured in cell lysates. As shown in Fig. 7B, CREB2/CaMKIV activation of the wild-type LTR ranged from 61.7-to 2201-fold. In comparison, CREB2/CaMKIV activation of the pLTR(E-box4-mutA)-luc and of the pLTR(E-box4-mutB)-luc ranged from 76.4-to 2656-fold, and from 95.1-to 2836-fold, respectively. Thus, in contrast to what we observed with Tax BLV , the E-box4-mutA and E-box4-mutB mutations did not reduce CREB2/CaMKIV responsiveness of the BLV LTR compared with the wild-type construct pLTRwt-luc. Similar results were observed in cotransfection experiments using either pSG-CREB2 alone, pCaMKIV alone, or after treatment with PMA/ ionomycin (data not shown).
Thus, our results demonstrate that mutations of the E-box4 motif located in the BLV R region impaired the responsiveness of the viral promoter to the viral protein Tax BLV but not to other activators known to up-regulate BLV expression.
The Effects of the Mutations in the E-box4 Motif Occur at the Transcriptional Level-In order to demonstrate that the amount of transcription (i.e. RNA levels) has been affected by the point mutations in the E-box4 motif, transcript levels in transiently transfected Raji cells were measured by RNase protection assays using probes proximal and distal to the BLV promoter (Fig. 8). The proximal probe, which overlaps the start of transcription in the BLV reporter plasmids, stretches from nt Ϫ118 to ϩ83 and therefore hybridizes to all transcripts that initiate at the BLV LTR to produce a protected species of 83 nt. The distal probe, producing a 225-nt protected luciferase product, can only detect transcripts that have extended into the luciferase gene and therefore provides a measure of elongation. We failed to observe any reporter transcripts in the absence of the transactivator Tax BLV with both the BLV promoter-specific probe and the luciferase gene-specific probe (data not shown), probably as a consequence of the weak BLV promoter activity in absence of Tax BLV and of the weak transfection efficiency of the DEAE-dextran procedure. Because Tax BLV is known to increase considerably the BLV LTR transcriptional activity and because the E-box4-mutA and E-box4-mutB mutations reduced both the basal and Tax BLV -activated BLV-driven gene expression (see top of Figs. 4 and 7A, respectively), we decided to analyze the RNA synthesis in the presence of Tax BLV to facilitate the evaluation of the luciferase gene expression directed by the wild-type and mutated LTRs. To this end, we performed RNase protection assays using RNAs extracted from Raji cells transiently cotransfected with either pLTRwt-luc, pLTR(E-box4-mutA)-luc, or pLTR(E-box4-mutB)-luc and 25 ng of the Tax BLV expression vector, pSG-TAX BLV . As expected, the E-box4-mutA and E-box4-mutB mutations reduced the luciferase activity by approximately 2-fold (Fig. 8A). In the same experiment, analysis of the steady-state mRNA level showed that these mutations also reduced transcript production, as detected with the proximal BLV promoter-specific probe or with the distal luciferase gene-specific probe (Fig. 8B, lanes 2  and 3). As an internal control, RNase protection analysis of the same RNA samples using an antisense probe corresponding to the GAPDH gene showed no change in the level of mRNA.
We thus demonstrate that the mutations A and B introduced in the E-box4 motif decrease the amount of transcription directed by the BLV promoter. These results are consistent with those of the LTR-luciferase assays and show that the effects of the E-box4 mutations occur at the level of transcription. IRF-2 in the first half of the BLV U5 region and have shown that mutation in this site causes a decrease in Tax-independent BLV LTR gene expression. Together with the E-box4 site described in this report, the IRF site in U5 constitutes the only two characterized transcription factor-binding sites located downstream of the BLV transcription start site. Therefore, double mutant constructs were generated in which both the E-box4 and the IRF site were mutated. The constructs pLTRwt-luc, pLTR(E-box4-mutA)-luc, and pLTR(E-box4-mutB)-luc were used as substrates to introduce a 3-bp mutation in the IRF motif. The three mutated plasmids were designated pLTR(IRFmut)-luc, pLTR(E-box4-mutA/IRFmut)-luc, and pLTR(E-box4-mutB/IRFmut)-luc, respectively. We tested the functional effects of the E-box4-mutA and E-box4-mutB mutations in combination with the mutation of the IRF-binding site by transient transfection of these plasmids into Raji cells. As shown in Fig. 9, mutation of the E-box4 motif resulted in a 22% reduction and a 35% reduction (depending on the mutation) of LTR-directed luciferase expression. Mutation of the IRF site resulted in a 42% reduction of LTR-directed luciferase expression. This result confirmed, in the context of the 344 BLV strain, the previous observation of our laboratory (22) showing that the IRF mutation induces a 2-fold reduction in basal transcription of an LTR isolated from the T15 BLV strain. Importantly, for the double mutants pLTR(E-box4-mutA/ IRFmut)-luc and pLTR(E-box4-mutB/IRFmut)-luc, we observed in a highly reproducible manner a 64 and 54% reduction, respectively, of LTR-directed luciferase expression (Fig. 9).

Mutations of Both the R Region E-box4 Site and the U5 Region IRF Site Decrease Basal Gene Expression from the BLV
Thus, we showed that mutation of the only two transcription factor-binding sites identified to date downstream of the transcription initiation site decreased the LTR basal activity more than 2-fold. These results reinforce the positive regulatory role played by the R region E-box4 motif and indicate that transcription factor-binding sites at the R/U5 junction are critical for optimal basal BLV promoter activity. DISCUSSION In this report, we have physically and functionally characterized an E box motif (referred to as E-box4) located in the R region of the BLV 5Ј-LTR. We have demonstrated by competition and supershift EMSAs that the bHLH transcription factors USF1 and USF2 bound in a sequence-specific manner to the E-box4-containing sequence. Mutations abolishing USF binding caused a reproducible decrease in BLV promoterdriven gene expression in transient transfection assays of Blymphoid cell lines. Cotransfection experiments showed that the USF1 and USF2a transactivators were able to act through the BLV R region E-box4. Moreover, combined mutation of both this E-box4 motif and the IRF site in U5 decreased the LTR basal activity to a greater degree than the individual mutations. This E box motif represents the first transcription factorbinding site reported in the BLV R region.
USF proteins belong to the bHLH-ZIP family of transcription factors, which includes c-Myc, Max (75), Mad (76), Mxi1 (77), AP4 (78), TFEB (29), TFE3 (27), MiTF (79,80), and ADD1 (81). USF was initially identified from HeLa cell nuclei and was shown to be necessary to stimulate transcription from the adenovirus major late promoter through the core sequence CACGTG (30, 60 -62). USF factors are encoded by two distinct genes (the USF1 and USF2 genes) (53,54,63,64,82). The USF1 gene encodes a single 43-kDa protein. The USF2 gene, due to alternative splicing, encodes two proteins, USF2a and USF2b of 44 and 38 kDa, respectively. The three USF isoforms differ in their N-terminal moieties, whereas they have highly conserved bHLH-ZIP domains (53,54). USF proteins form homo-and heterodimers both in vitro and in vivo (53,82). Dimerization with other bHLH proteins has not been observed (27,33,75,76,83). Although expression of the USF1 and USF2 species is ubiquitous, different ratios of USF homo-and heterodimers are found in different cell types (53). The major USF species present in most tissues and cell lines is the USF1-USF2 heterodimer. USF1 homodimers are less abundant, and USF2 homodimers are usually quite scarce (53,54). USF proteins are ubiquitous transcription factors that were initially considered to play a role in housekeeping functions (61). Furthermore, these USF proteins have also been recognized as important players in the transcriptional regulation of tissue-specific genes (26, 84 -88) and in the specific response of genes to external modulators (89), as glucose (46,90,91). The mechanisms by which USF proteins contribute to these specific functions are yet unclear and may overlay several phenomena. USF-binding sites have been found, and the involvement of USF in transcriptional regulation has been studied in a number of cellular and viral genes (46,90,(92)(93)(94)(95).
In this report, we clearly demonstrated a significant decrease in BLV promoter activity accompanying mutation in the E-box4, indicating a positive functional role of this motif. Moreover, cotransfection experiments using USF1 and USF2a expression vectors demonstrated the ability of a DNA fragment corresponding to multiple E-box4 copies to confer USF inducibility to a minimal TK promoter in an E-box4-dependent manner. We also showed that ectopic USF1 and USF2a transcription factors up-regulated BLV promoter activity. This USF responsiveness of the BLV promoter was mediated in part through the E-box4 motif. However, the three other E boxes located in U3 were also responsible for part of the USF response. We are currently investigating the identity of the transcription factors binding to these three E boxes. Interestingly, we observed that mutation of all four E boxes did not com- pletely attenuate the stimulatory effect of USF1 and USF2a in cotransfection experiments. Rather, there was a reduction of stimulation to ϳ50% that observed with the wild-type BLV promoter reporter construct. The presence of other yet unidentified E boxes in the BLV LTR could be a first explanation. However, computer analysis searches of the complete BLV LTR did not reveal any other CANNTG consensus sequence (data not shown), suggesting that non-E box cis-elements within the BLV promoter are directly or indirectly responsive to USF. A report by Desbarats et al. (96) documented a similar situation with the rat prothymosin-␣ intron enhancer that contains a critical E box; ectopic expression of USF was shown to equally transactivate reporter plasmids carrying this sequence with the E box either intact or mutated, indicating that activation did not occur entirely through the E box element. Similarly, activation by USF has been reported for a number of other promoters either lacking an E box or containing a mutated E box (93,97,98). Our results are consistent with these previous studies and suggest that USF may interact directly and indirectly with non-E box cis-elements in the BLV promoter region. Alternatively, overexpression of USF may be modulating the level or activity of other proteins or basal transcription factors that mediate the activity of the BLV promoter.
Studies by other investigators (60) have reported that USF is able to bind sequences other than CACGTG such as the initiator element in the adenovirus major late promoter. Moreover, some studies have shown that USF factors can stimulate transcription by direct interaction with members of the basal initiation complex as follows: the basal factor TFIID (60, 63), the initiator-binding protein TFII-I (99, 100), and TBP-associated factor TAFII55 (101). USF proteins have also been shown to interact with other transcription factors, such as Fra1 (102), c-Maf (103), Ets1 (104), and Sp1/Sp3 (105). Thus, substantial evidence indicates that USF is involved in regulating gene expression by direct binding to E boxes and/or initiator elements and by functionally interacting with basal and/or specific transcription factors other than USF. The present study showed that USF regulated the expression of BLV in part through four E boxes (the E-box-1-3 in U3 and the E-box4 in R). The residual USF induction observed with the LTR construct mutated in all four E boxes could result from USF action directly or indirectly through other cis-regulatory elements. Additional experiments will be necessary to test this hypothesis.
Our in vitro binding studies demonstrated that both USF1 and USF2 specifically interacted with the E-box4 motif. Our ex vivo functional studies showed that ectopic USF1 and USF2a proteins activated transcription of a reporter gene driven by either the homologous BLV promoter or a minimal heterologous promoter containing multiple upstream E-box4 motifs. In these latter studies, both USF1 and USF2a exerted their stimulatory effect with similar efficiency. Although from different genes, USF1 and USF2 display a high degree of homology in their C-terminal DNA binding domain and USF-specific region that is thought to play a role in transcriptional activation of promoters (53,106). Moreover, studies with USF null mice (55) and specific genes such as L-type pyruvate kinase (54), liver fatty-acid synthase (66), and human dipeptidyl peptidase IV (107) have shown that USF1 and USF2 have overlapping, redundant activities and can transactivate some promoters with similar efficiency. Therefore, our results are consistent with these previous observations, because they demonstrate that the BLV LTR is another promoter activated equally well by USF1 or USF2a, suggesting that both USF genes are involved in BLV transcriptional activation.
The members of the bHLH transcription factor family have been divided into four classes (A-D) depending on the sequence of the canonical bHLH-binding site CANNTG (25). Class B proteins, which include c-Myc, Max, MyoD, myogenin, and USF, bind to CACGTG or CATGTG. The E-box4 motif identified here in the BLV R region is an E box class B consensus motif and accordingly was found to bind the class B proteins USF1 and USF2. We could not observe the binding of c-Myc to this motif by supershift experiments, even though Myc/Max heterodimers share the same binding site requirements as USF dimers. Interestingly, in vivo, the myc oncogene is overexpressed in B-lymphocytes from tumors (108) and PBMCs isolated from BLV-infected animals with persistent lymphocytosis (109). Moreover, it has been reported that levels of Myc/Max proteins in nuclear extracts are very difficult to detect, whereas USF is the major binding activity detected by in vitro assays with crude nuclear extracts from several cell types and species (98, 110 -114), even from cells that are transformed by c-Myc (112). For all these reasons, although we did not observe Myc binding under our in vitro conditions, we considered the possibility that, under physiological conditions, the E-box4 motif would be nevertheless permissive for Myc interactions. Toward this end, we decided to analyze the consequences of c-Myc overexpression on the activity of the BLV promoter. We cloned the ovine c-Myc cDNA (43) in the eukaryotic expression vector pCDNA3, and we performed transient cotransfection experiments into both B-cell lines (Raji and Daudi) and non-B-cell lines (HeLa) of this Myc expression construct with the LTR BLVluciferase vector in the absence or presence of a Max expression vector. Our results did not reveal any stimulatory effect of Myc/Max on the BLV promoter. 2 Therefore, our data strongly suggest that USF proteins but not Myc-Max complexes are the functional transcription factors that can activate the E-box4 in the R region of BLV. However, it should be stressed that we did not investigate the factors that bind to the E-box4 motif in vivo, and therefore the binding of factors different from both USF and Myc cannot be excluded at this time.
Our data showed that the E-box4 positively regulated BLV LTR-directed gene expression both in the absence and presence of Tax BLV ex vivo and could therefore be involved in the early and late stages of viral infection, respectively. USF proteins, in conjunction with CREB, ATF-1, ATF-2 (18,19), IRF-1, and IRF2 (22), could thus be important transcription factors involved in the initiation of BLV transcription. Indeed, the E-box4 motif may with others initiate a low level of transcription from the BLV LTR promoter and lead to the synthesis of small amounts of the Tax BLV transactivator, which could then amplify transcription of the viral genome. Although a 64-bp DAS at the 3Ј end of the R region (ϩ147 to ϩ211) was reported previously (23), no transcription factor-binding site had been identified in or close to the DAS. The E-box4 motif described in this report is located right in the center of DAS and could therefore be, at least in part, responsible for the positive regulatory activity of DAS. Moreover, we have demonstrated that mutation of both the R region E-box4 site and the U5 region IRF site decreased Tax BLV -independent LTR-driven gene expression to a greater degree than mutation of either site separately. These results indicate that the two transcription factorbinding sites identified to date downstream of the transcription start site play a critical role in the basal transcriptional regulation of the BLV promoter. Downstream regulatory sequences have also been identified in the HTLV-I LTR. Indeed, a 45-bp element, which is located at the boundary of R-U5 and binds the YB-1 transcription factor, is required for Tax HTLV-I -independent transcription (115,116). On the other hand, binding of the Sp1 and Sp3 transcription factors to the HTLV-1 U5 region has been associated with transcriptional repression of the LTR (117,118). Furthermore, it has been suggested that the interaction of CREB and ATF-2 with the R region of the HTLV-1 LTR is associated with viral latency (119,120).
We have shown that the E-box4 motif also positively regulates BLV LTR-directed gene expression in the presence of Tax BLV and could therefore be involved in the late stage of viral replication. How USF affects the responsiveness of the BLV LTR to Tax BLV and whether this effect is mediated by direct or indirect interactions of USF with Tax BLV remains to be established. Interestingly, the cellular transcription factor USF has been reported to cooperate with other viral transactivators, such as the E2 proteins of human papillomavirus type 16 and bovine papillomavirus type 1 (121), and the immediate-early protein 62 of varicella-zoster virus (93). USF has been demonstrated to interact with the adenovirus E1A protein, which stimulates transcription of adenovirus genes as well as a wide variety of other viral and cellular genes (122). Moreover, a few studies show that the Tax protein of HTLV-1 represses the expression of cellular genes through the bHLH transcription factors. First, Uittenbogaard et al. (123,124) have demonstrated that members of the bHLH protein family mediate repression by the HTLV-1 Tax protein, including the transcriptional repression of the p53 gene. They show that this repression of p53 by Tax HTLV-1 is dependent upon the p53 promoter E box element and is mediated by the class B bHLH proteins c-Myc and USF (124). Second, Tax HTLV-1 has been shown to mask c-Myc function through a c-AMP-dependent pathway (125). The region in c-Myc perturbed by Tax HTLV-1 is contained within a highly conserved transformation/activation domain. Their data provide evidence that the N-terminal portion of c-Myc is conformationally altered by Tax HTLV-1 , without additional effects on overall protein stability. A conformational disruption of this protein could account for an abrogation of the transactivation and transformation properties of Myc (125). Third, the biological behavior of p56 lck is modulated by the expression of the viral regulatory tax gene in HTLV-1-infected T-cells through a mechanism of repression that involves the E box DNA recognition sequence encountered in the lck gene distal promoter (126). Fourth, a molecular mechanism for Tax HTLV1 -mediated repression of the transcriptional activity of the bHLH myogenic MyoD protein has been proposed recently (127). In this study, the authors show that Tax HTLV-1 binding to the KIX domain of the cellular coactivator p300 prevents MyoD from contacting this N-terminal domain of p300, thus resulting in repression of MyoD-dependent transcription (127). Regarding the results reported in the present study, it would be interesting to study the possibility that Tax BLV may inhibit the transcriptional activation function of bHLH proteins, thereby interfering with cellular gene expression and/or BLV expression.
We have characterized a novel E box motif in the R region of the BLV LTR. Functionally important E box motifs have been identified in the regulatory regions of other viruses. Latencyassociated promoter 1 (LAP1) of herpes simplex virus type 1 is required to generate a series of latency-associated transcripts in sensory neurons of latently infected animals, and a USF element and a CRE site contribute to LAP1 function during latency (128). The transcriptionally regulatory regions of the lymphomagenic Akv and SL3-3 murine leukemia retroviruses contain E box motifs that are important determinants for murine leukemia retrovirus transcriptional activity in lymphocytic cell lines (129). The murine sarcoma virus enhancer has six E box targets for MyoD family proteins (130). Activation of the Epstein-Barr virus (EBV) DNA polymerase promoter by the BRLF1 immediate-early viral transactivator is mediated through USF and E2F (131). The latent membrane protein 1 (LMP1) gene promoter in the EBV genome contains a silencing activity overlapping with a transcriptional enhancer in a region that contains an E box motif (71). The latter study shows that bHLH transcription factors Max, Mad1, USF, E12, and E47 and the corepressor mSin3A bind in vitro to this E box (71). The silencing and enhancer activities correlate with the binding of Max-Mad1 and of USF, respectively (71). Finally, four E boxes are present in the promoter region of human immunodeficiency virus type 1 (HIV-1) (51,52,(132)(133)(134)(135), and cooperative interaction of USF1 with Ets-1 is required for full transcriptional activity of the HIV-1 LTR in T-cells (104).
In summary, we provided evidence that the BLV R region E-box4 is required, and USFs are components for BLV promoter regulation by correlating functional assays and USF binding activities to the E-box4. Our findings contribute toward a clearer understanding of the transcriptional regulation of BLV. Further studies will focus on characterizing the physiological role of the R region E-box4 site in the context of the complete infectious BLV provirus. A BLV provirus mutated in this site will be constructed. This mutated provirus will be injected into sheep in order to test for infectivity and ability to induce pathogenesis. These in vivo experiments should provide new insight into the molecular mechanisms of BLV transcriptional activation.