An ATF/CRE Element Mediates both EBNA2-Dependent and EBNA2-Independent Activation of the Epstein-Barr Virus LMP1 Gene Promoter

Abstract

The Epstein-Barr virus (EBV) latent membrane protein 1 (LMP1) is a viral oncogene whose expression is regulated by both viral and cellular factors. EBV nuclear antigen 2 (EBNA2) is a potent transactivator of LMP1 expression in human B cells, and several EBNA2 response elements have been identified in the promoter regulatory sequence (LRS). We have previously shown that an activating transcription factor/cyclic AMP response element (ATF/CRE) site in LRS is involved in EBNA2 responsiveness. We now establish the importance of the ATF/CRE element by mutational analysis and show that both EBNA2-dependent activation and EBNA2-independent activation of the promoter occur via this site but are mediated by separate sets of factors. An electrophoretic mobility shift assay (EMSA) with specific antibodies showed that the ATF-1, CREB-1, ATF-2 and c-Jun factors bind to the site as ATF-1/CREB-1 and ATF-2/c-Jun heterodimers whereas the Sp1 and Sp3 factors bind to an adjacent Sp site. Overexpression of ATF-1 and CREB-1 in the cells by expression vectors demonstrated that homodimeric as well as heterodimeric forms of the factors transactivate the LMP1 promoter in an EBNA2-independent manner. The homodimers of ATF-2 and c-Jun did not significantly stimulate promoter activity. In contrast, the ATF-2/c-Jun heterodimer had only a minor stimulatory effect in the absence of EBNA2 but induced a strong transactivation of the LMP1 promoter when coexpressed with this protein. Evidence for a direct interaction between the ATF-2/c-Jun heterodimeric complex and EBNA2 was obtained by EMSA and coimmunoprecipitation experiments. Thus, our results suggest that EBNA2-induced transactivation via the ATF/CRE site occurs through a direct contact between EBNA2 and an ATF-2/c-Jun heterodimer. EBNA2-independent promoter activation via this site, on the other hand, is mediated by a heterodimeric complex between the ATF-1 and CREB-1 factors.

Epstein-Barr virus (EBV) is a ubiquitous human herpesvirus, consistently detected in several human malignancies, including endemic Burkitt’s lymphoma (BL), nasopharyngeal carcinoma (NPC), and posttransplantation lymphoma (50). In vitro infection of B lymphocytes by EBV, as well as explant culture of lymphocytes from seropositive adults, gives rise to immortalized cell lines with the limited gene expression of six nuclear proteins (EBNA1 to EBNA6) and three membrane proteins (LMP1, LMP2A, and LMP2B), as well as two small nuclear RNAs (EBER1 and EBER2) (36). Mutagenesis of the viral genome has defined a subset of six genes required for (EBNA1 to EBNA3, EBNA6, and LMP1) or contributing to (EBNA5) B-cell immortalization (12, 30, 35, 45, 58, 70).

The EBNA2 protein transactivates the LMP1 gene but also transactivates other viral and cellular genes (1, 13, 19, 24, 38, 57, 63–67, 71). However, since EBNA2 seems to lack sequence-specific DNA-binding ability, the participation of cellular proteins is necessary for the recognition of specific promoters. Some of these proteins have been identified, including C promoter binding factor 1 (CBF1), also designated Jκ recombination signal-binding protein (RBP-Jκ) (26, 31, 43, 62, 72), the Ets-related PU.1 factor (33), and a POU domain protein (55). It has been suggested that EBNA2, when bound by cellular proteins, associates with specific regulatory sites in viral or cellular genomes and activates transcription by recruiting basal transcription factors to nearby promoters. This is corroborated by the recent demonstration of a direct interaction between EBNA2 and components of the RNA polymerase II transcription initiation complex (59–61, 68).

Several lines of evidence indicate that the transforming effect of LMP1 is explained to a large extent by its functional similarity to an activated form of tumor necrosis factor family receptor (TNFR). This notion is based on the facts that LMP1 has an intrinsic ability to aggregate in the plasma membrane and to associate with TNFR-associated factors (17, 42, 44, 48). The expression of the LMP1 gene in B cells is primarily due to activation of a promoter designated EDL1 in the EBV genome (22, 23). In the present study, we have focused on the promoter-proximal part of the LMP1 regulatory sequence (LRS) which contains a potential Sp site at position −33 and an activating transcription factor/cyclic AMP (cAMP) response element (ATF/CRE) site at position −41 relative to the transcription initiation site (see Fig. 1). The Sp factor-binding element is one of the most widely distributed promoter elements in cellular and viral genes. To date, four different Sp proteins, designated Sp1, Sp2, Sp3, and Sp4, have been identified. The members of this transcription factor family have structural features in common, including zinc fingers and glutamine- and serine/threonine-rich amino acid stretches (27, 37). A previous study showed that the ATF/CRE motif in LRS is likely to play a role as a mediator of the EBNA2 effect and promoter activation by cAMP analogs (21). The ATF/CRE sequence motif belongs to one of the major classes of regulatory elements that participates in transcriptional regulation induced by extracellular signals. Several proteins, including ATF-1, ATF-2, ATF-3, ATF-a, CREB-1, CREB-2, and the CREM proteins, bind as homo- or heterodimers to this sequence (15). This family of regulatory factors has been implicated in cAMP-, calcium-, and virus-induced modulation of transcription (15, 25, 54). The AP-1 binding site (TRE), which confers responsiveness to tetradecanoyl phorbol acetate (2) differs by only 1 nucleotide from the ATF/CRE site. This element is recognized by a group of proteins, including those encoded by the c-fos and c-jun gene families, that form homo- and heterodimers with each other (29). The members of the AP-1 and ATF/CREB factor families bind preferentially to their respective sequence, but due to selective formation of interfamily heterodimers, new binding specificities arise. For example, c-Jun binds to an ATF/CRE site with considerably higher affinity as a heterodimer with ATF-2 than as a c-Jun homodimer (29).

FIG. 1.

Schematic presentation of the LRS in the B95-8 EBV genome. The scale refers to the position relative to the transcription initiation site from the EDL1 promoter. Transcription factor-binding sites previously identified as involved in regulation of the LMP1 promoter are indicated by open boxes. The Sp and the ATF/CRE sites are defined in the present investigation.

The objective of the present study was to define the role of the Sp and ATF/CRE sites in EBNA2 responsiveness of the LMP1 promoter and to characterize the factors involved. Mutational analysis showed that both elements were required for an efficient response of the promoter. Electrophoretic mobility shift assay (EMSA) and antibody supershift analysis demonstrated that Sp1 and Sp3 bound to the Sp site and that two distinct heterodimeric complexes, ATF-1/CREB-1 and c-Jun/ATF-2, interacted with the ATF/CRE site. The results indicate that the EBNA2 transactivation of the promoter is dependent on interaction with the c-Jun/ATF-2 heterodimer whereas the previously shown stimulatory effect of cAMP on LMP1 expression probably is mediated by the ATF-1/CREB-1 heterodimer.

MATERIALS AND METHODS

Plasmids.

All constructs made were verified by dideoxy sequencing utilizing the Sequenase system (United States Biochemical Corp., Cleveland, Ohio). The pSV2gpt, pEΔA6, pIBI31(BYRF), pgCAT, pgLRS(−54)CAT, pgLRS(−106)CAT, and pgLRS(−634)CAT constructs have been described previously (20, 52, 55). The LRS is defined as nucleotides 169477 to 170151 of B95-8 EBV DNA, which corresponds to positions −634 to +40 relative to the transcription initiation site.

To make a series of Sp and ATF/CRE mutated reporter plasmids, PCR amplifications were performed with the pgLRS(−152)CAT plasmid (55) as a template and primers that resulted in fragments with one end corresponding to position +40 in LRS and the other end corresponding to position −58, with mutations in the Sp site (G to T in position −33 and −32) or the ATF/CRE site (C to A in position −40 and G to T in position −41). The PCR fragments were cloned into the TA cloning vector (Invitrogen, NV Leek, The Netherlands). Taking advantage of a synthetic HindIII site in one primer and a PstI site in the TA cloning vector, the PCR fragments were then cloned between the HindIII and PstI sites in the pgCAT plasmid. To generate the pgLRS(−106)CAT plasmid with the Sp or ATF/CRE site mutated, the pgLRS(−58)CAT constructs were digested with HindIII and MluI and the HindIII-MluI LRS fragments that contained the −54/+40 part of LRS were isolated. The wild-type pgLRS(−106)CAT was cleaved with the same enzymes, and the MluI-HindIII fragment corresponding to positions −106 to −55 was isolated and ligated with the mutated −54/+40 LRS fragments and HindIII-cleaved pgCAT, generating pgLRS(−106)(Spmut)CAT and pgLRS(−106)(ATF/CREmut)CAT. The mutated pgLRS(−634)CAT constructs were generated by ligating the mutation-containing HindIII-MluI fragments described above into the HindIII-MluI-digested pgLRS(−634)CAT vector, resulting in pgLRS(−634)(Spmut)CAT and pgLRS(−634)(ATF/CREmut)CAT. The pgLRS(−106)ΔCAT vector used for in vitro transcription of the probe in RNase protection experiments was made by cloning the PvuII-SalI LRS fragment of pgLRS(−106)CAT in the SmaI-SalI sites of the Gemini 3Zf(+) vector (Promega Corp., Madison, Wis.). The pCMV-Sp1 plasmid was a gift from G. Suske (Klinikum der Philipps-Universität Marburg, Marburg, Germany). The cDNA for human ATF-2 was kindly provided by C. Svensson-Akusjärvi (Uppsala University, Uppsala, Sweden). The ATF-2-encoding cDNA was amplified by PCR and cloned into the TA cloning vector. The ATF-2 gene-containing fragment was excised with BstXI and ligated into the pcDNAI/Amp plasmid (Invitrogen), resulting in the pc(ATF-2) expression vector. An expression vector for human ATF-1, designated pc(ATF-1), was made by cloning the human cDNA for ATF-1 obtained by XbaI cleavage of the pET-15b(ATF-1) plasmid, a gift from R. H. Goodman (Oregon Health Science University, Portland, Oreg.), in pc(ATF-1). An expression vector for rat CREB-1, designated pc(CREB-1), was generated by cloning the CREB341 cDNA-containing BamHI-XbaI fragment of pET-15b(CREB341), also provided by R. H. Goodman, in pcDNAI/Amp. To generate an expression vector for human c-Jun, cDNA was isolated from the pCMV c-jun vector, kindly supplied by R. Tjian (University of California Berkeley, Berkeley, Calif.) by cleavage with BamHI and PvuII and was cloned into BamHI-EcoRV-digested pcDNAI/Amp. The resulting plasmid was designated pc(c-jun).

The EBNA2 expression vector pc(BYRF) was constructed as follows. First, part of the BYRF open reading frame was subjected to PCR amplification with a sense oligonucleotide with two new restriction enzyme sites (XhoI-NdeI) and the BYRF translation start sequence and an antisense primer corresponding to the sequence around the BamHI Y/H cleavage site. Then the fragment was excised and subcloned between the XhoI and BamHI sites in pEΔA8 (52), re-creating the complete BYRF sequence with the addition of a NdeI site close to the translation initiation codon. The EBNA2-encoding NdeI-BglII fragment of this plasmid was cloned in the XbaI site of the pCI vector (Promega) with XbaI linkers. Finally, the EBNA2-encoding EcoRI-SalI fragment of this plasmid was cloned between the EcoRI-XhoI sites in the pcDNAI/Amp plasmid, creating the pc(BYRF) vector. The pE300CY6 vector, which allows the expression of a truncated version of rat CD2, was most kindly provided by E. Lundgren (University of Umeå, Umeå, Sweden), and the E1A 13S expression vector was provided by C. Svensson-Akusjärvi.

Cell culture, DNA transfections, and CAT assays.

DG75 is an EBV genome-negative BL cell line (6). The lymphoid cells were maintained as suspension cultures in RPMI 1640 medium (Life Technologies AB, Täby, Sweden) supplemented with 10% fetal calf serum (Life Technologies AB), penicillin, and streptomycin. Transfections were generally performed with 5 × 10⁶ DG75 cells, 6.0 to 10 μg of DNA of the reporter construct to be tested, and 1.4 pmol of DNA of the EBNA2 expression vector pEΔA6 or the pSV2gpt control plasmid by electroporation at 260 V and 960 μF in 250 μl of cell culture medium with the Bio-Rad Gene Pulser and 4-mm-gap cuvettes (Bio-Rad, Hercules, Calif.). The cells were harvested after 72 h, and aliquots of the cell lysates were assayed for chloramphenicol acetyltransferase (CAT) activity (51). For Sp1 transfections (see Fig. 6), 0.95 pmol of the pCMV-Sp1 vector or the pCMV control vector, 0.68 pmol of the EBNA2 expression vector pEΔA6 or the pSV2gpt control, and 6.0 μg of the reporter plasmid were used. In the ATF-1 and CREB-1 transfections (see Fig. 7A), 3.6 pmol of either pc(ATF-1) or pc(CREB-1) or, alternatively, 1.8 pmol of each of the vectors or 3.6 pmol of the pcDNAI/Amp (from now on designated pc) control vector were used together with 0.68 pmol of EBNA2 expression vector pEΔA6 or 0.68 pmol of the pSV2gpt control vector and 5.0 μg of the respective reporter construct. In the ATF-2 and c-Jun transfections (see Fig. 7B), 3.6 pmol of pc(ATF-2), or 0.20 pmol of pc(c-jun) and 3.4 pmol of pc, or 1.8 pmol of pc(ATF-2) and 0.10 pmol of pc(c-jun) and 1.7 pmol of pc, or 3.6 pmol of pc was used together with 0.68 pmol of EBNA2 expression vector pc(BYRF) or 0.68 pmol of the pc control vector and 5.0 μg of the respective reporter construct. For the study of the effect of EBNA2 on phosphorylation (see Fig. 8), 1.4 pmol of pEΔA6, pSV2gpt, and the E1A 13S expression vector, respectively, was cotransfected with 10 μg of the pE300CY6 plasmid into 10⁷ DG75 cells. In the immunoprecipitation experiments (see Fig. 10) 1.8 pmol of pc(ATF-2) and 0.1 pmol of pc(c-jun) were cotransfected with 10 μg of the pE300CY6 plasmid into 10⁷ DG75 cells together with either 0.68 pmol of the EBNA2 expression vector pc(BYRF) or 0.68 pmol of the pc control vector.

FIG. 6.

Sp1 transactivates the LMP1 promoter independently of EBNA2. The pCMV-Sp1 expression vector or an equivalent amount of the empty pCMV control plasmid was cotransfected with the EBNA2 expression vector pEΔA6 or the pSV2gpt plasmid with the pgLRS(−106)CAT reporter plasmid or the mutated derivative pgLRS(−106)(Spmut)CAT in DG75 cells. The CAT activity is expressed as percent chloramphenicol acetylation, with the value obtained with the pCMV plasmid together with pEΔA6 and pgLRS(−106)CAT as 100%. The standard errors are indicated by error bars. The 100% value corresponded to 22% conversion of substrate to product in the CAT assay. The values presented are the mean of four independent transfections.

FIG. 7.

The LMP1 promoter can be transactivated by CREB-1 and ATF-1 homo- and heterodimers independently of EBNA2 and by a c-Jun/ATF-2 heterodimer in an EBNA2-dependent manner. (A) The pc(ATF-1), and pc(CREB-1) expression vectors, separately or mixed, or the pc control plasmid was cotransfected with the EBNA2 expression vector pEΔA6 or an equivalent amount of pSV2gpt and the reporter plasmids pgLRS(−106)CAT or pgLRS(−106)(ATF/CREmut)CAT into DG75 cells, as detailed in Materials and Methods. The CAT activity is expressed as the percent chloramphenicol acetylation relative to the value obtained in transfections with the pc plasmid together with pEΔA6 and pgLRS(−106)CAT. The 100% value corresponded to 21% conversion of substrate to product in the CAT assay. The standard errors are indicated with error bars. The values shown are the mean of three independent transfections. (B) The pc(ATF-2) and pc(c-Jun) expression vectors, separately or in combination, or the pc control plasmid was cotransfected with the EBNA2 expression vector pc(BYRF) or an equivalent amount of the pc plasmid and the reporter plasmids pgLRS(−106)CAT or pgLRS(−106)(ATF/CREmut)CAT into DG75 cells, as detailed in Materials and Methods. The CAT activity is expressed as the percent chloramphenicol acetylation relative to the value obtained in transfections with the pc plasmid together with pc(BYRF) and pLRS(−106)CAT. The 100% value corresponded to 12% conversion of substrate to product in the CAT assay. The standard errors are indicated by error bars. The values shown are the mean of three independent transfections.

FIG. 8.

EBNA2 does not affect the level or phosphorylation state of c-Jun or ATF-2 in DG75 cells. The EBNA2 expression vector pEΔA6 or equivalent amounts of the pSV2gpt control plasmid or the E1A 13S expression vector were transfected together with the CD2 expression vector pE300CY6 in DG75 cells. The transfected cells were selected for their CD2 expression with magnetic beads. The cells were lysed and equal amounts of protein extract were analyzed by SDS-PAGE and immunoblotting. The antibodies used in panel A were anti-c-Jun, anti-phospho-c-Jun(Ser63), and anti-phospho-c-Jun(Ser73), and those used in panel B were anti-ATF-2 and anti-phospho-ATF-2(Thr71). NIH 3T3 cell extracts containing nonphosphorylated or phosphorylated forms of c-Jun (panel A, lanes 1 and 2) and ATF-2 (panel B, lanes 1 and 2), respectively, were used as controls of antibody activity. The anti-c-Jun(Ser73) antibody also detected JunD phosphorylated at the Ser100 residue.

FIG. 10.

EBNA2 coimmunoprecipitates with the c-Jun/ATF-2 heterodimer. The EBNA2 expression vector pc(BYRF) or equivalent amounts of the pc control plasmid were transfected together with pc(c-Jun) and pc(ATF-2) and the CD2 expression vector pE300CY6 in DG75 cells. The transfected cells were selected for CD2 expression with magnetic beads. After lysis of the cells, the proteins were immunoprecipitated with specific antibodies and adsorption to protein A/G agarose, eluted, and analyzed by SDS-PAGE and immunoblotting. Cell extracts that had not been subjected to immunoprecipitation were analyzed in lanes 1 to 3, and control immunoprecipitations without the specific antibody but including the adsorption step with protein A/G were analyzed in lanes 10 to 12. Lanes: 1, DG75 cells transfected with pc(BYRF); 2, DG75 cells transfected with the pc plasmid; 3, B95-8 cells; 4, anti-ATF-2 precipitate from DG75 cells transfected with pc(BYRF); 5, anti-ATF-2 precipitate from DG75 cells transfected with the pc plasmid; 6, anti-ATF-2 precipitate from B95-8 cells; 7, anti-c-Jun precipitate from DG75 cells transfected with pc(BYRF); 8, anti-c-Jun precipitate from DG75 cells transfected with the pc plasmid; 9, anti-c-Jun precipitate from B95-8 cells; 10, protein A/G agarose eluate from DG75 cells transfected with pc(BYRF); 11, protein A/G agarose eluate from DG75 cells transfected with the pc plasmid; 12, protein A/G agarose eluate from B95-8 cells. Antibodies used for visualizing the proteins on the immunoblots were anti-ATF-2 (A), anti-c-Jun (B), and a human serum containing anti-EBNA2 antibodies (C). The positions of ATF-2, c-Jun, EBNA2, and immunoglobulin heavy chains (Ig H) are indicated by the solid arrowheads.

RNase protection assay.

Cytoplasmic RNA was prepared and analyzed by the RNase protection assay as described previously (51). ³²P-labelled RNA was synthesized by in vitro transcription of pgLRS(−106)ΔCAT with [α-³²P]UTP (3000 Ci/mmol; Du Medical Scandinavia AB, Sollentuna, Sweden) and T7 RNA polymerase by following a standard procedure. Hybridization was performed at 50°C.

EMSA.

Nuclear extracts were prepared as described previously (18), except that antipain (5.0 μg/ml), leupeptin (5.0 μg/ml), and aprotinin (2.0 μg/ml) were added to the buffer in the final homogenization and dialysis steps and phenylmethylsulfonyl fluoride was substituted by Pefabloc (0.50 mM). Aliquots were frozen in liquid nitrogen and stored at −70°C. The following double-stranded synthetic oligonucleotides were used in the mobility shift assays: an oligonucleotide corresponding to the −50 to −19 part of LRS; a similar oligonucleotide in which the Sp site was mutated by the introduction of G-to-T mutations between nucleotides −33 and −32; and an oligonucleotide corresponding to an AP-1 consensus sequence (5′-CGCTTGATGACTCAGCCGGAA-3′). The blunt-ended oligonucleotides were labelled with [γ-³²P]ATP (6,000 Ci/mmol, Du Medical Scandinavia AB) by using polynucleotide kinase (Boehringer Mannheim Scandinavia AB, Bromma, Sweden). The labelled probes were purified by electrophoresis in a 5% polyacrylamide gel (acrylamide/bisacrylamide, 30:1) in 0.5× TBE (50 mM Tris, 50 mM boric acid, 1.0 mM EDTA [pH 8.3]). The wet gel was autoradiographed, and the DNA fragments were excised, electroeluted by isotachophoresis (48a), and precipitated. Binding-reaction mixtures (in 25 μl) with crude nuclear extract contained 10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 1.0 mM dithiothreitol, 1.0 mM EDTA, 5% glycerol, various amounts (0.30 to 7.0 μg) of poly(dA-dT), 6.0 fmol of [³²P]DNA (approximately 70,000 cpm) and various amounts (1.0 to 20 μg) of nuclear proteins (always added last). Binding-reaction mixtures (in 25 μl) with in vitro-translated proteins contained 10 mM HEPES (pH 7.9), 50 mM KCl, 6.0 mM MgCl₂, 2.5 mM dithiothreitol, 100 μg of bovine serum albumin (BSA) per ml, 0.01% Nonidet P-40, 10% glycerol, 6.0 fmol of [³²P]DNA (approximately 70,000 cpm), and 5.0 μl of programmed rabbit reticulocyte lysate. In the competition experiment, a 200- or 300-fold excess of competing oligonucleotide was added before the ³²P-labelled probe. After incubation at room temperature for 25 min, the samples were electrophoresed on 5% polyacrylamide gels (acrylamide/bisacrylamide, 30:1) in 0.5× TBE. The unlabelled competitors used were as follows: probe oligonucleotide corresponding to LRS −50/−19, 5′-GAGGCTTATGTAGGGCGGCTACGTCAGAGTAA-3′; nonspecific oligonucleotide, 5′-ATGTTCGGTAACATCTCTCATTGCGCACAAAGAACCCTACATCCG-3′; ATF/CRE consensus oligonucleotide, 5′-AAGATTGCCTGACGTCAGAGAGCTAG-3′; Sp consensus oligonucleotide, 5′-ATTCGATCGGGGCGGGGCGAGC-3′; The HindIII-MluI fragment of pgLRS(−106)(Spmut)CAT corresponding to LRS −54 to +40 with a mutated Sp site (two G nucleotides at positions −32 and −33 were replaced by two T nucleotides); the corresponding HindIII-MluI fragment of pgLRS(−106)(ATF/CREmut)CAT with a mutated ATF/CRE site (CG at positions −40 and −41 were replaced by AT); the corresponding HindIII-MluI fragment of pgLRS(−106)(Sp+ATF/CREmut)CAT with mutated Sp and ATF/CRE sites (C, T, G, and A at positions −37, −38, −41, −44, were changed to T, A, A, and T).

The antibodies against the transcription factors Sp1 (sc-59X), Sp3 (sc-644X), CREB-1 (sc-271X), CREB-2 (sc-200X), ATF-1 (sc-243X), ATF-2 (sc-187X), c-Jun (sc-822X), c-Jun/Jun B/Jun D (sc-44X), Jun B (sc-46X), and Jun D (sc-74X) (Santa Cruz Biotechnology, Inc., Santa Cruz, Calif.) were used. The supershift analyses were performed as described above for the EMSA experiments except that 3.0 to 8.0 μl of the respective antibody was added after the incubation at room temperature. The mixture was incubated at 4°C for 60 min and then subjected to polyacrylamide gel electrophoresis (PAGE) (5.0% polyacrylamide) followed by autoradiography.

For in vitro expression, a fragment of pEΔA6 containing the EBNA2-encoding open reading frame, BYRF1, was subcloned into the pIBI 31 vector (IBI, New Haven, Conn.). The cDNAs for ATF-2 and c-Jun were translated in vitro with the pc(ATF-2) and pc(c-jun) constructs. The supercoiled DNA templates were sequentially transcribed and translated in the same reaction mixture containing rabbit reticulocyte lysate (Promega Corp.), amino acids, and the T7 RNA polymerase, as recommended by the manufacturer. Translated proteins were analyzed by sodium dodecyl sulfate (SDS)-PAGE.

CD2 selection.

To introduce a marker for cell sorting, the CD2 expression vector pE300CY6 was cotransfected with the EBNA2 expression vector (pEΔA6), the E1A 13S expression vector, and the pSV2gpt control plasmid. The transfected cells were collected by centrifugation after 48 h and washed. Sorting for CD2 expressing cells was performed with a mouse CD2-specific antibody (MCA154; Serotec, Oxford, United Kingdom) and magnetic beads linked to rat anti-mouse antibodies (Dynabeads M450; Dynal Ltd., Merseyside, United Kingdom) as described by Pilon et al. (49). The cells were lysed in lysis buffer (20 mM imidazole-HCl [pH 6.8], 100 mM KCl, 1.0 mM MgCl₂, 10 mM EGTA, 0.20% [vol/vol] Triton X-100, 10 mM NaF, 1.0 mM sodium vanadate, 1.0 mM sodium molybdate, 5.0 μg of leupeptin per ml, 2.0 μg of aprotinin per ml), incubated for 15 min at 4°C, and cleared by centrifugation. The protein concentration in the lysates was determined (Bradford protein assay; Bio-Rad), and aliquots were mixed with sample buffer containing 65 mM Tris-HCl (pH 6.8), 2.0% SDS, 10% glycerol, 5.0% β-mercaptoethanol, and 0.10% bromphenol blue (BPB). A 50-μg portion of protein from the respective sample was boiled and subjected to SDS-PAGE (10% polyacrylamide). Total c-Jun, c-Jun phosphorylated at Ser63 or Ser73, total ATF-2, and ATF-2 phosphorylated at Thr71 were detected with the PhosphoPlus antibody kits (New England Biolabs, Inc., Beverly, Mass.). It should be noted that the anti-c-Jun(Ser73) antibody also recognizes JunD phosphorylated at Ser100. The negative control consisted of total-cell extract from NIH 3T3 cells, and the positive controls were extracts of UV-treated (c-Jun) or anisomycin-treated (ATF-2) NIH 3T3 cells.

Immunoprecipitation and immunoblot analysis.

DG75 cells were harvested 48 h after transfection and subjected to CD2 selection as described above. B95-8 cells and the selected DG75 cells were lysed as described above, sonicated, and cleared by centrifugation. Aliquots corresponding to 0.9 × 10⁶ DG75 cells and 1.8 × 10⁶ B95-8 cells were immunoprecipitated with 30 μg of anti-ATF-2 (sc-187X; Santa Cruz Biotechnology, Inc.) per ml and 30 μg of anti-c-Jun (sc-822X; Santa Cruz Biotechnology, Inc.) per ml, respectively, in a total volume of 300 μl and incubated at 4°C overnight. Samples without antibody were used as negative controls. Aliquots (50 μl) of 50% protein A/G agarose (Santa Cruz Biotechnology, Inc.) in lysis buffer containing 10 μg of BSA per ml were added, the samples were rocked at 4°C for 2 h, and the protein A/G agarose beads were collected by centrifugation. After the beads had been washed five times in lysis buffer plus BSA, the proteins were eluted by boiling in 40 μl of sample buffer, analyzed by SDS-PAGE (10% polyacrylamide), and blotted to Hybond C-extra nitrocellulose membranes (Amersham Life Science, Little Chalfont, United Kingdom). The membranes were incubated with rabbit anti-c-Jun antibodies (sc-822X), mouse anti-ATF-2 antibodies (sc-187X), or a human serum containing anti-EBNA2 antibodies in phosphate-buffered saline (PBS; 180 mM NaCl, 3.6 mM KCl, 11 mM Na₂HPO₄, 2.0 mM KH₂PO₄) containing 0.5% nonfat dry milk and, after repeated washings in PBS, incubated with horseradish peroxidase-conjugated donkey anti-rabbit (Amersham Life Science), sheep anti-mouse (Amersham Life Science), or goat anti-human (Bio-Rad) antibodies. The membrane was washed in PBS containing 0.3% Tween 20, and the proteins were visualized by enhanced chemiluminescence procedures as described by the manufacturer of the reagents (Amersham Life Science).

RESULTS

EBNA2-induced transactivation of the LMP1 promoter depends on intact Sp and ATF/CRE motifs in LRS.

In previous studies, we have demonstrated that the −106 to +40 part of LRS contains one or several elements that mediate EBNA2-induced upregulation of promoter activity (19, 21, 55). Two short subsequences in this region were defined with homology to an Sp and an ATF/CRE site, respectively (Fig. 1) (21). Mutation analysis provided evidence for a role of the ATF/CRE site in promoter transactivation (21). However, subsequent binding studies revealed that this mutation prevented the binding of factors to both the Sp site and the ATF/CRE site. To assess the relative contribution of the two sites to promoter activity and to further elucidate their role in the EBNA2-induced transactivation process, LRS-carrying reporter plasmids were created with specific mutations of the respective sites. The pgLRS(−106)CAT plasmid contained the −106 to +40 part of LRS and was mutated as indicated in Materials and Methods. To determine the effect of the mutations in the context of the complete LRS, mutated derivatives of the pgLRS(−634)CAT reporter plasmid were generated. The plasmids were cotransfected with an EBNA2 expression vector or a control plasmid into the EBV-negative DG75 B-cell line. Mutation of the ATF/CRE site in pgLRS(−106)CAT reduced the EBNA2-induced activity to 8.0% relative to the wild-type plasmid, i.e., close to the activity of the background plasmid pgCAT (Fig. 2). Mutation of the Sp site left a low residual EBNA2-induced activity of 16% relative to the wild-type plasmid (Fig. 2). The corresponding mutations in the LRS(−634)CAT plasmids similarly resulted in a pronounced reduction of activity (Fig. 2). Thus, the results clearly showed that the two sites are important for the EBNA2-dependent transactivation of the LMP1 promoter, especially in the context of the promoter-proximal part of LRS. The full-length LRS seemed to contain elements that to a certain extent could compensate for the loss of the ATF/CRE site at position −41.

FIG. 2.

EBNA2-induced transactivation of LRS depends on intact Sp and ATF/CRE motifs in LRS. Mutations were introduced in the pgLRS(−106)CAT and pgLRS(−634)CAT plasmids, respectively, as indicated in Materials and Methods. The reporter plasmids were cotransfected with pEΔA6 (+EBNA2) or with an equivalent amount of pSV2gpt (−EBNA2) in the EBV-negative B-cell line DG75. The CAT activity is given as relative chloramphenicol acetylation expressed as a percentage of the activity obtained with pgLRS(−634)CAT in the presence of EBNA2. The 100% value corresponded to acetylation of 97% of the substrate in the assay. The values are the mean of three independent transfections. The standard errors are indicated by error bars.

To confirm that the observed transactivation of the reporter plasmids was due to correctly initiated transcripts from the EDL1 promoter, RNase protection analysis was performed. The result showed that transcription was initiated at the correct LRS position in the constructs investigated (Fig. 3).

FIG. 3.

EBNA2-induced transcription initiates at the correct LMP1 promoter site in reporter plasmids. RNA was prepared from DG75 cells transfected with the EBNA2 expression vector pEΔA6 or the pSV2gpt control vector and the indicated LRS CAT reporter plasmids and subjected to RNase protection analysis with a ³²P-labelled probe corresponding to positions −106 to +40 of LRS and the first part of the CAT gene. Lanes: 1, probe only; 2, pgLRS(−54)CAT and pSV2gpt; 3, pgLRS(−54)CAT and pEΔA6; 4, pgLRS(−106)CAT and pSV2gpt; 5, pgLRS(−106)CAT and pEΔA6; 6, pgLRS(−634)CAT and pSV2gpt; 7, pgLRS(−634)CAT and pEΔA6; 8, pgCAT and pSV2gpt; 9, pgCAT and pEΔA6; 10, DNA size markers. The band corresponding to the common initiation site in the EDL1 promoter is indicated by the solid arrow. Bands corresponding to nonspecific initiation upstream of LRS in the vector part of the reporter plasmids are indicated by dotted arrows. The lengths of these protected fragments differ depending on the plasmid. The solid arrowhead indicates a band present in all samples, probably due to incomplete RNase cleavage.

Identification of factors binding to the Sp and ATF/CRE motifs.

We next characterized the regulatory factors in DG75 cells that bound to the Sp and ATF/CRE elements by performing EMSAs with a double-stranded oligonucleotide corresponding to the −50 to −19 part of LRS. Competition experiments with unlabelled oligonucleotides containing intact Sp or ATF/CRE consensus sequences or LRS fragments with mutations of the corresponding sites were carried out to correlate the resulting EMSA bands with the binding sites. Five specific complexes were identified (Fig. 4, lanes 2 and 3). Bands that were not abolished by competition with unlabelled probe were assumed to represent nonspecific complex formation. It might be noted that the same binding pattern was observed both with EBV-negative and EBV-positive B cells and with epithelial cells and T cells (data not shown). Competition with an LRS fragment that contained a mutated Sp site (lanes 6 and 7) removed three of the bands and left two bands. These complexes presumably represented binding to the Sp site. Competition with an LRS fragment that contained a mutated ATF/CRE site (lanes 10 and 11) removed the two bands marked Sp and left three bands. These were assumed to represent binding to the ATF/CRE site. Furthermore, an LRS fragment mutated in both the Sp and the ATF/CRE site did not compete with any of the complexes (lanes 8 and 9). Competition with an oligonucleotide that contained an Sp consensus sequence removed the two Sp bands (lane 5), and an oligonucleotide that contained an ATF/CRE consensus sequence removed the three ATF/CRE bands (lane 4). It was noted that although the EMSA band patterns were similar in qualitative terms, the intensity of the bands was weaker when DNA fragments (lanes 6 to 11) were used as competitors compared with the bands obtained with the corresponding synthetic oligonucleotides (lanes 3 to 5), even though similar molar amounts of competitor had been used. The reason is not clear, but we suggest that the effect was nonspecific and was related to the fact that competitors of different lengths were used (the average length of the oligonucleotides was about 30 bp, and the length of the LRS fragments was about 90 bp).

FIG. 4.

Sp and ATF/CRE transcription factors in B-lymphoid cells bind to LRS. A ³²P-labelled double-stranded synthetic oligonucleotide corresponding to the −50 to −19 LRS region was incubated with nuclear extracts from DG75 cells and subjected to EMSA. Lane 1 shows the binding pattern obtained with the nuclear extract. Competition reactions was carried out as indicated below the autoradiogram and described in Materials and Methods. In lanes 2 to 5, the binding mixtures contain a 300-fold excess of unlabelled competitor over probe; in lanes 6, 8, and 10, they contain a 200-fold excess; and in lanes 7, 9, and 11, they contain a 300-fold excess. Some of the competitors were mutated at the Sp and/or the ATF/CRE sites as specified in Materials and Methods. Five complexes indicated by solid arrows are considered specific and designated Sp (bands remaining after competition with an LRS fragment that contained a mutated Sp site) and ATF/CRE (bands remaining after competition with an LRS fragment that contained a mutated ATF/CRE site), respectively. Three nonspecific bands that were not abolished by competition with unlabelled probe are indicated by dotted arrows.

To identify the members of the Sp and ATF/CREB transcription factor families that were involved in the formation of a complex with the −50/−19 LRS probe, we performed antibody supershift analysis with different commercially available antibodies (Fig. 5). To simplify the band pattern, the first series of EMSAs were carried out as competition experiments with a 300-fold molar excess of the unlabelled LRS probe containing a mutated Sp binding site, which allowed only Sp-related factors to bind to the labelled probe. As illustrated in Fig. 5A, one of the complexes supershifted with an anti-Sp1 antibody (lane 2) and the other two supershifted with an anti-Sp3 antibody (lanes 3 and 4). One of the Sp3-containing complexes was hidden behind the strong band corresponding to the Sp1 complex; therefore, the supershift became evident only when the anti-Sp1 and anti-Sp3 antibodies were added simultaneously. The anti-Sp3 antibody removed the two Sp3-containing EMSA complexes but did not give rise to supershifted bands in the gel, due to the inhibition of complex formation by the antibody.

FIG. 5.

Identification of the transcription factors interacting with the Sp and ATF/CRE motifs in LRS. (A) Nuclear extract of DG75 cells was incubated under binding conditions with a ³²P-labelled double-stranded oligonucleotide corresponding to the −50 to −19 LRS region in the presence of a 300-fold molar excess of the competitor LRS −50/−19 with a mutated Sp site. Antibody supershifts were carried out by incubation with a goat polyclonal antibody against Sp1, a rabbit polyclonal antibody against Sp3, and a mixture of the antibodies, as indicated below the autoradiogram. The reaction mixtures were analyzed by EMSA. Three specific complexes are indicated by solid arrows, one designated Sp1 and two designated Sp3. Two bands that were not abolished by competition are indicated by the dotted arrows. The position of the anti-Sp1 antibody-shifted complex is shown by the solid arrowhead. (B and C) EMSA and antibody supershift analysis were performed by incubating nuclear extract of DG75 cells under binding conditions with a ³²P-labelled double-stranded oligonucleotide corresponding to the LRS −50 to −19 region with a mutated Sp site and with antibodies as indicated below the autoradiogram. Two bands that were not abolished by competition are indicated by the dotted arrows. (B) Three specific complexes are indicated by solid arrows, two of which are designated ATF-1, CREB-1 since they contain both factors. The positions of the antibody complexes are indicated by an open arrowhead for the anti-CREB-1 shift and solid arrowheads for the anti-ATF-1 shifts. (C) The third of the three specific complexes indicated by solid arrows is identified as ATF-2, c-Jun. The positions of the immunologically shifted complexes are shown by the solid arrowheads for the anti-ATF-1 shifts and the open arrowhead for the anti-c-Jun shift.

The ATF/CREB antibody supershift analyses were carried out with a labelled −50/−19 LRS probe with a mutated Sp site, which allowed the formation of complexes only with the ATF/CRE site of the probe (Fig. 5B and C). Two of the three complexes were supershifted by both an anti-CREB-1 antibody (Fig. 5B, lane 2) and an anti-ATF-1 antibody (lane 4). The third ATF/CRE complex was removed by an anti-ATF-2 antibody (Fig. 5C, lane 3) and an anti-c-Jun antibody (lane 5) and shifted by another anti-c-Jun antibody (lane 4).

Involvement of an Sp site in the regulation of the LMP1 promoter.

Our results showed that the Sp site at position −33 in the promoter-proximal region of LRS had to be present to achieve EBNA2-induced transactivation of the LMP1 promoter and that both the Sp1 and Sp3 factors bound to this site. The ability of Sp1 to transactivate the promoter was assessed by transient transfections with Sp1 and EBNA2 expression vectors and a pgLRS(−106)CAT reporter plasmid. In the absence of EBNA2, the Sp1 vector induced a low level of transactivation compared with the control plasmid (Fig. 6). The activation was, however, significantly higher than that obtained with a reporter plasmid in which the Sp site was mutated. The Sp1 expression vector did not add to the activity of the pgLRS(−106)CAT plasmid induced by EBNA2, although mutation of the Sp site largely abolished promoter activity. This suggests that Sp1 has an EBNA2-independent stimulatory effect on LMP1 promoter activity. Protein levels in the transfected cells were checked by immunoblot analysis (data not shown). The amount of Sp1 in the cells increased from a basal level after transfection with the expression vector.

ATF-1 and CREB-1 can activate the LMP1 promoter in an EBNA2-independent manner.

We have previously reported on the existence of a possible relationship between the EBNA2 effect on LMP1 in B cells and the cAMP signal transduction pathway and provided evidence for the notion that the ATF/CRE site in the −106/+40 part of LRS is a possible target in the EBNA2-induced activation of the promoter (21). The EMSA results of the present study indicated that the CREB-1 and ATF-1 factors are candidate mediators of the activating effect. To investigate this question, ATF-1 and CREB-1 expression vectors, separately or in combination, were cotransfected with the pgLRS(−106)CAT plasmid with or without an EBNA2 expression vector in DG75 cells. As illustrated in Fig. 7A, overexpression of CREB-1 or ATF-1 activated the LRS(−106)-containing reporter plasmid independently of EBNA2. The activating effect was largely abolished when the ATF/CRE site was mutated. The residual ATF-1-mediated transactivation of the mutated LRS reporter plasmid most probably did not originate from the ATF/CRE site, since EMSA analysis showed that the transcription factors no longer bound to the mutated site (data not shown). Transfection with a mixture of half the amount of each of the expression vectors resulted in a level of activation intermediate between those obtained with the ATF-1 and a CREB-1 expression vectors separately. Concomitant expression of EBNA2 did not significantly increase the activity of the LRS reporter plasmid. Protein levels in the transfected cells were checked by immunoblot analysis (data not shown). The ATF-1 and CREB-1 protein concentrations increased from a basal level, and the EBNA2 protein appeared when the respective expression vectors were introduced into the cells. We suggest that the CREB-1 and ATF-1 factors separately and together are able to activate the LMP1 promoter via the ATF/CRE site at position −41 in LRS independently of each other and of EBNA2.

EBNA2 can transactivate the LMP1 promoter via the ATF/CRE site and a c-Jun/ATF-2 heterodimer.

According to the EMSA results, the ATF-2 and c-Jun factors interact at the ATF/CRE site at position −41 in the LRS. To investigate the possible role of these factors in the EBNA2-dependent activation of the EDL1 promoter, ATF-2 and c-Jun expression vectors were transfected together with an EBNA2 expression vector and pgLRS(−106)CAT in DG75 cells. In the absence of EBNA2, the homodimers of ATF-2 or c-Jun did not significantly increase the basal activity of the reporter plasmid (Fig. 7B). The heterodimeric forms of the factors transactivated the promoter about twofold. However, coexpression of ATF-2, c-Jun, and EBNA2 in the cells resulted in a strong and ATF/CRE site-dependent activation of the promoter. Immunoblotting control experiments showed that ATF-2 and c-Jun protein levels increased and that EBNA2 appeared in the cell extracts after transfection (data not shown). It should be noted that a cytomegalovirus promoter-driven EBNA2 expression vector was used in these experiments since the EBNA2 expression of our standard pEΔA6 plasmid was downregulated in the presence of c-Jun. This new vector expressed EBNA2 less well, giving rise to seemingly lower EBNA2 inducibility of the reporter constructs.

It has been established that phosphorylation of defined amino acid residues in c-Jun (Ser-63 and Ser-73) and ATF-2 (Thr-69 and Thr-71) is required to generate efficient transactivational function of the factors. Therefore, the following question arises: does EBNA2 modify the phosphorylation state and/or the levels of these factors as a means of inducing activation of the LMP1 promoter? This possibility was studied in EBNA2 cotransfection experiments with commercially available antibodies with the ability to specifically identify phosphorylated forms of the c-Jun and ATF-2 proteins (Fig. 8). Induction of phosphorylation and increased expression of c-Jun by the adenovirus E1A protein was used as an experimental control. The results showed that EBNA2 did not significantly affect either the total level or the phosphorylation state at the Ser-63/Ser-73 and Thr-71 residues, respectively, of c-Jun or ATF-2. The DG75 cells apparently contained a significant endogenous level of the factors in phosphorylated form. Thus, the results support the notion that EBNA2 transactivation of the LMP1 promoter does not occur through the phosphorylation of c-Jun or ATF-2.

To correlate the observed effects of overexpression of ATF-2 and/or c-Jun on promoter activity with the binding of the respective factor at the ATF/CRE site and to decide whether EBNA2 interacts with any of these factors, a series of EMSAs were performed with in vitro translation reaction mixtures containing ATF-2, c-Jun, and EBNA2, respectively. Notably, the bands corresponding to the homomeric forms of in vitro-synthesized ATF-2 and c-Jun were not affected by the addition of EBNA2 (Fig. 9A, lane 5, Fig. 9B, lane 4). However, the heterodimeric complex formed by in vitro-translated ATF-2 and c-Jun was removed by the addition of EBNA2 to the EMSA reaction mixture, showing that EBNA2 specifically interacts with the heterodimer but not with the homodimeric forms of the two transcription factors (Fig. 9C, lane 6). The reason why the in vitro-translated heterodimer of c-Jun/ATF-2 migrated more slowly than the in vivo-synthesized heterodimer in the nuclear extracts is not clear. However, the results were also corroborated by an EMSA experiment where addition of in vitro-translated EBNA2 to the nuclear extract removed the c-Jun/ATF-2 complex from the EMSA pattern but left the ATF-1/CREB-1 factors bound to the probe (data not shown).

FIG. 9.

In vitro-translated EBNA2 abrogates the binding of the in vitro-translated heterodimer c-Jun/ATF-2, but not the respective homodimeric forms, to the ATF/CRE site. The ³²P-labelled oligonucleotide probes indicated in the figure were incubated with DG75 nuclear extract and/or in vitro-translated proteins and analyzed by EMSA. The three specific complexes obtained with DG75 nuclear extract and identified above are indicated by solid arrows and designated ATF-1, CREB-1 and ATF-2, c-Jun. In vitro-translated proteins are denoted by the prefix IVT, and the positions of the corresponding complexes are shown by solid arrows. Nonspecific bands that were not abolished by competition are indicated by the dotted arrows. (A) The binding-reaction mixtures contained a ³²P-labelled −50 to −19 LRS oligonucleotide probe with a mutated Sp site and DG75 nuclear extract (lane 1) or the same probe with in vitro-translated ATF-2 (lanes 2 to 6). Antibody supershifts were performed by incubation with a rabbit polyclonal antibody against ATF-2 (lane 3) or a rabbit polyclonal antibody against CREB-2 (lane 4). A supershifted band is indicated by a solid arrowhead. In lanes 5 and 6, aliquots of reticulocyte in vitro translation reactions with EBNA2 DNA or control DNA were added to the binding-reaction mixtures. (B) The binding-reaction mixtures contained a ³²P-labelled AP-1 consensus oligonucleotide probe and in vitro-translated c-Jun protein (lane 1). Antibody supershifts were performed by incubation with a mouse monoclonal antibody against c-Jun (lane 2) or a goat polyclonal antibody against Jun B (lane 3). A supershifted band is indicated by a solid arrowhead. In lanes 4 and 5, aliquots of reticulocyte in vitro translation reaction mixtures with EBNA2 DNA or control DNA were added to the binding-reaction mixtures. (C) The binding-reaction mixtures contained ³²P-labelled −50 to −19 LRS oligonucleotide probe with a mutated Sp site and DG75 nuclear extract (lane 1) or in vitro-translated ATF-2 and c-Jun protein (lanes 2 to 7). Antibody supershifts were performed by incubation with a rabbit polyclonal antibody against ATF-2 (lane 3), a mouse monoclonal antibody against c-Jun (lane 4), or a goat polyclonal antibody against Jun B (lane 5). Supershifted bands are indicated by solid and open arrowheads. In lanes 6 and 7, aliquots of reticulocyte in vitro translation reaction mixtures with EBNA2 DNA or control DNA were added to the binding-reaction mixtures.

The results of the EMSA analysis suggested that EBNA2 has the ability to interact with the c-Jun/ATF-2 heterodimer. To corroborate this observation, experiments aimed at identifying a complex between EBNA2 and c-Jun and ATF-2 in cell extracts by immunoprecipitation were performed (Fig. 10). Anti-ATF-2 and Anti-c-Jun antibodies were used to precipitate the putative complex, the precipitates were analyzed by SDS-PAGE and immunoblotting, and the proteins were identified with anti-ATF-2 (Fig. 10A), anti-c-Jun (Fig. 10B), and anti-EBNA2 (Fig. 10C) antibodies. The results showed that ATF-2 and c-Jun were precipitated by the respective antibody and that EBNA2 coprecipitated with both ATF-2 and c-Jun (Fig. 10C, lanes 4, 6, 7, and 9). In the controls without the primary antibody (lanes 10 to 12), none of the proteins were detected. The recombinant c-Jun protein had a somewhat lower molecular weight than endogenously expressed c-Jun, but the reason for this is not clear. It should be noted that the EBNA2/c-Jun/ATF-2 complex was identified not only in a situation of overexpression of the respective factors but also in nontransfected, EBV-infected cells (B95-8 cells). Together, the results of the EMSA and the immunoprecipitation experiments strongly suggest that EBNA2 can transactivate the LMP1 promoter via the ATF/CRE motif and that a direct interaction between EBNA2 and a heterodimeric complex of c-Jun and ATF-2 constitutes an essential step in the activation process.

DISCUSSION

We have shown in previous studies that EBNA2-induced transactivation of the LMP1 promoter in lymphoid cells depends to a significant extent on transcriptional cis-elements in the −106/+40 part of LRS and that an ATF/CRE site in this region most probably plays a role as a mediator of the EBNA2 effect and in promoter activation by cAMP analogs (21, 55). In this study, using the EBV-negative DG75 cell line as a model system for B cells, we demonstrated that the LMP1 promoter can be activated in an EBNA2-independent manner via a process that includes the binding of ATF-1 and CREB-1 homo- or heterodimers to the ATF/CRE site whereas EBNA2-dependent activation of the promoter, on the other hand, occurs through a pathway that involves a direct interaction between EBNA2 and ATF-2–c-Jun heterodimers at the same site. In addition, the activity of the promoter is modulated in an EBNA2-independent way by the interaction of Sp1 and possibly Sp3 with an adjacent Sp element. We have previously compared the activity of the LRS in a variety of cell lines of epithelial and B-cell origin, including DG75 (20). That study was not as detailed as the present one, but an important conclusion was that all group I BL cell lines investigated, both EBV negative and EBV positive, followed the same pattern with regard to EBNA2-induced transactivation of LRS. We therefore consider the DG75 cell line to be representative of its category of B cells and useful for investigations of the regulation of the LMP1 promoter. The reason why several other investigations have failed to detect EBNA2 responsiveness in the proximal LMP1 promoter region is not clear to us. It has been suggested that a cryptic RBP-Jκ site in our reporter constructs is involved. A search for structural motifs in the proximal part of the LRS with reasonable identity to a RBP-Jκ site revealed the presence of a TTGGGAT sequence at positions −67 to −61. However, the results of EMSA competition analysis and site-directed mutagenesis strongly argued against the possibility that RBP-Jκ or any other factor binds to this motif (unpublished results). Furthermore, we have consistently obtained the same EBNA2-induced promoter response with the −106/+40 LRS fragment inserted in an unrelated plasmid carrying the luciferase reporter gene. The EBNA2 expression vector used in our previous studies is a genomic construct that would hypothetically allow the synthesis of a mini-version of EBNA5 or some other unidentified product and might in this way be responsible for discrepancies between our observations and those of others. To eliminate this possibility, we repeated the experiments with another expression vector in which the cytomegalovirus early promoter was placed immediately upstream of a DNA fragment containing only the EBNA2-encoding BYRF1 open reading frame from the B95-8 EBV strain. This construct was considerably less efficient in EBNA2 expression than was our standard EBNA2 expression vector pEΔA6 but produced the same result in qualitative terms with regard to EBNA2 responsiveness of the −106/+40 LRS region (data not shown). Furthermore, we have never detected peptide material reacting with anti-EBNA5 antibodies in EBV-negative cells (DG75 and COS-1 cells) transfected with the pEΔA6 plasmid by immunoblot or immunofluorescence analysis (data not shown).

EBNA2-independent activation of LRS via the Sp and the ATF/CRE sites.

Mutational analysis and EMSA binding studies showed that an Sp element at position −33 in LRS is required for efficient EBNA2-dependent and EBNA2-independent transactivation of the LMP1 promoter and that the Sp1 and Sp3 transcription factors bind to this site. Sp1 is a well-known transcriptional activator. The limited effect of overexpression of Sp1 on promoter activity in the absence of EBNA2 obtained in our transfection experiments might be explained by the fact that DG75 cells contain a high endogenous level of the Sp1 factor that diminishes the relative contribution of the exogenously added protein. The Sp3 transcription factor has been shown to function as a repressor of Sp1-mediated transcriptional activation (28). Multiple Sp3-containing complexes similar to those observed in our EMSA analyses have previously been found in another system (16). We suggest that the stimulatory effect of Sp1 on the LMP1 promoter is independent of EBNA2 but is a prerequisite for EBNA2 induced transactivation. The interaction of the Sp1 and Sp3 factors with their binding site in LRS might constitute an EBNA2-independent regulatory system in which the balance between the positively acting Sp1 and the negatively acting Sp3 factors is one of the factors that determines the final level of activity of the LMP1 promoter.

It is now generally believed that bZIP proteins like ATF-1 and CREB-1 bind to DNA only as dimers and not as monomers. The results of the EMSA and antibody supershift experiments in this study suggested that the ATF-1 and CREB-1 proteins in DG75 cells bind to the ATF/CRE site at position −41 in LRS as a heterodimer, since the homomeric forms of the factors were not detected. The presence of two ATF-1/CREB-1 complexes with different mobilities in the electrophoretograms is explained by the previous observation that phosphorylation drastically changes the conformation of ATF-1 and, as a consequence, the electrophoretic mobility of the corresponding EMSA complex (46). It should be noted, however, that overexpression of the factors in the cells by transfection with expression vectors under conditions that favored the formation of the homodimeric forms showed that these were as efficient in inducing promoter activity as was the heterodimeric form (Fig. 7A).

The ATF-1 and CREB-1 factors are phosphorylated by protein kinase A (PKA), which in most cases appears indispensable for activation (for a review, see reference 47). The major effect of phosphorylation seems to occur at the level of the transactivating function, while the effects on dimerization and DNA binding are less certain. Protein phosphatase 1 (PP-1) dephosphorylates ATF-1 and CREB-1 and correspondingly attenuates the transactivational activity of the factors. The phosphatase inhibitor protein-1 (IP-1) is a specific inhibitor of PP-1, and its activity is dependent on phosphorylation by PKA. Thus, activation of PKA by cAMP would result in the phosphorylation and activation of ATF-1/CREB-1 and IP-1, with the latter leading to the inhibition of PP-1. Studies of purified ATF/CREB proteins have demonstrated that the negatively charged and phosphorylated kinase-inducible domain of the factors is responsible for the interaction with components of the basal transcription apparatus. We have previously suggested a model for EBNA2-induced transactivation of the LMP1 promoter that involved the ATF/CRE site and a direct inhibition of PP-1-catalyzed dephosphorylation of CREB-1 by EBNA2 in an IP-1-analogous manner (21). However, our present investigation does not support the hypothesis that activation of the ATF/CRE site would occur through an EBNA2-induced increase of the phosphorylated form of CREB-1 or ATF-1 at the binding site. EBNA2 did not significantly increase the stimulatory effect of ATF-1 and CREB-1 on LRS-CAT reporter plasmids in DG75 cells (Fig. 7A) or change the phosphorylation status of the transcription factors (data not shown).

The importance of this ATF/CRE site in LRS is also emphasised by the results of Chen et al. (7). They showed that a sequence variant found in the corresponding ATF/CRE motif of an NPC EBV isolate, when transferred into a B95-8 LRS sequence, conferred a threefold reduction of the activity of the LMP1 promoter both in B cells and in epithelial cells. Interestingly, the NPC sequence variant diminished the absolute levels of activity of a reporter plasmid that carried the −495/+20 part of LRS in both the absence and presence of EBNA2 but did not change the relative level of EBNA2 responsiveness in B cells. This indicates that the EBNA2-independent promoter activation pathway is disrupted by this sequence variant. It is consistent with our conclusion that the ATF/CRE site is one of the limiting factors that determine the final level of activity of the LMP1 promoter under different induction conditions and cellular environments.

Treatment of BL lines in the EBNA1-positive form of latency (latency I) with anti-immunoglobulin or with tetradecanoyl phorbol acetate induces the lytic cycle via the protein kinase C (PKC) signal transduction pathway. The switch to the lytic cycle involves the rapid upregulation of LMP1 in the cells and occurs independently of EBNA2 expression (53). It has been shown that CREB-1 and possibly ATF-1 can be activated by phosphorylation via the PKC pathway in B lymphocytes (69). It is thus conceivable that the LMP1 promoter can be activated through the PKC pathway as well as the PKA pathway and that the signal to the general transcriptional machinery is mediated in both cases by ATF-1/CREB-1 dimers bound to the ATF/CRE site.

EBNA2-dependent activation of LRS via the ATF/CRE site.

The results of this study lend strong support to the notion that EBNA2 can activate the LMP1 promoter via a mechanism that is different from the ATF-1/CREB-1 pathway discussed above and that involves the binding of the ATF-2 and c-Jun factors as a heterodimer to the ATF/CRE site. EBNA2 is required for the activation (Fig. 7B) and, judging from the EMSA (Fig. 9C) and coimmunoprecipitation (Fig. 10) experiments, seems to make a direct contact with the c-Jun/ATF-2 dimer complex. Thus, the question arises of how this interaction may lead to promoter activation. Does EBNA2 induce a modification of the ATF-2/c-Jun dimer and/or its binding site or change the concentration of the factors in the cell nucleus in a way that favors promoter activation through the activating domains of ATF-2 and c-Jun? Or is EBNA2 recruited to the LMP1 promoter by protein-protein interactions with the ATF-2/c-Jun dimer to bring the EBNA2 transactivational domain in the correct position for a productive contact with one or several distinct general transcription factors? The possibility also exists that the interaction between EBNA2 and the c-Jun/ATF-2 dimer decreases the affinity of this complex for the ATF/CRE site, leading to an increased binding of the ATF-1 and CREB-1 factors and activation of the LMP1 promoter through this pathway. The fact that overexpression of ATF-2 and c-Jun in the presence of EBNA2 has a pronounced activating effect on the LMP1 promoter (Fig. 7B) strongly argues against such a hypothesis. With regard to the first alternative, we have not been able to detect any change in the phosphorylation status or the levels of ATF-2 and c-Jun in parallel with the EBNA2-induced activation of the LMP1 promoter (Fig. 8). In addition, it has been demonstrated in several studies that the C-terminal acidic domain of EBNA2 is required for transcriptional transactivation by EBNA2 (10, 11, 56), and the activating domain of EBNA2 has been found to make physical contact with several general transcription factors, including TFIIB, TAF40, and TFIIH (59, 61). Thus, it seems quite likely that EBNA2, at least in the context of the −106/+40 part of LRS, functions in a manner analogous, in several respects, to the transcriptional coactivators CBP (CREB-binding protein) and the adenovirus E1A-associated cellular protein p300 with regard to the ATF/CRE. Neither CBP nor p300 by itself binds to DNA, but they can be recruited to promoter elements by interaction with a multitude of sequence-specific activators. These interactions include CBP-CREB (9, 39), p300-CREB (3), CBP–c-Jun (4), p300-YY1 (41), CBP-Fos (5), CBP–c-Myb (14), and CBP-nuclear receptors (34). CBP can activate transcription through a glutamine-rich region in the C-terminal part of the protein, and the activation domain has been shown to interact with components of the basal transcription machinery (39). Thus, CBP and p300 are transcriptional coactivators that provide a crucial link between transcriptional activators stimulated by signalling cascades and initiation of transcription. EBNA2 seems to function through a similar mechanism.

EBNA2 interacts with several other transcriptional regulatory elements in the LMP1 promoter. These factors include the RBP-Jκ (26, 31, 43, 62, 72), the Ets-related PU.1 factor (33), and an unidentified member of the POU domain-containing protein family (55). RBP-Jκ is a transcriptional repressor which binds to DNA sequences (GTGGGAA) in LRS. It has been shown that EBNA2 can act by targeting DNA-bound RBP-Jκ within the nucleus and abolishing RBP-Jκ-mediated repression through masking of the repression domain (26, 31, 32, 43, 62, 72). Furthermore, EBNA2 interactions with PU.1 and the POU domain protein seem to be essential for the efficient upregulation of the LMP1 promoter, and the elements might act in a cooperative manner (33, 55). The number of EBNA2 molecules bound to the LMP1 promoter in its activated configuration is not known. However, in a simplistic model, Johannsen et al. (33) have suggested that one EBNA2 molecule (presumably as a dimer) binds to the regulatory region via multiple protein-protein interactions with a number of transcriptional activators including PU.1, LBF3, LBF5, LBF6, LBF7, and RBP-Jκ. In the light of our investigations, we would suggest that the c-Jun/ATF-2 heterodimer and the POU domain protein should also be included. The biochemical function of these multiple contact points would then be to increase the stability of the promoter-EBNA2 complex and hence the specificity and efficiency of the induction. Functional studies are consistent with the notion that some of the activators surrounding the PU.1-binding site cooperate with PU.1 in the binding of the same EBNA2 dimer (33, 40, 55). Independent binding of separate EBNA2 molecules to multiple sites in vivo through different factors including RBP-Jκ, LBF3, LBF5, LBF6, LBF7, PU.1, the POU domain protein, and the c-Jun/ATF-2 heterodimer is, however, also possible, although we have no data to support such an assumption. It has recently been demonstrated in a model system that multiple EBV ZEBRA molecules bound upstream of the TATA box and initiation site synergistically interact with TFIID and TFIIA, resulting in the assembly of a preinitiation subcomplex (the DA complex) and a concomitant isomerization (8). Once isomerized, the complex binds TFIIB and the remaining general factors. Interestingly, the recruitment of the DA complex required multiple contacts and is therefore the basis for transcriptional synergy in the system. Recruitment and isomerization of the DA complex may be a general effect of many activators and coactivators, including EBNA2.