Activation of the Epstein-Barr virus transcription factor BZLF1 by 12-O-tetradecanoylphorbol-13-acetate-induced phosphorylation

Abstract

BZLF1 is a member of the extended AP-1 family of transcription factors which binds to specific BZLF1 sequence motifs within early Epstein-Barr virus (EBV) promoters and to closely related AP-1 motifs. BZLF1's activity is regulated at the transcriptional level as well as through protein interactions and posttranslational modifications. Phorbol esters or immunoglobulin cross-linking both reactivate EBV from latently infected B cells via transactivation of BZLF1. We report here that the phorbol ester 12-O-tetradecanoylphorbol-13-acetate (TPA) is capable of inducing BZLF1's activity even further. The induction occurs at the posttranscriptional level and depends on a single serine residue located in the DNA binding domain of BZLF1. This serine residue (S186) is phosphorylated by protein kinase C in vitro and in vivo after stimulation with TPA. Phosphorylation of S186 per se interferes with the DNA binding affinity of BZLF1 in vitro but is mandatory for TPA-induced increase in DNA binding of BZLF1, as shown in gel retardation assays and reconstruction experiments with cellular extracts. In transcriptional reporter assays, S186 is essential for the activation of BZLF1 by TPA. Presumably, a yet-to-be-identified cellular factor restores the DNA binding affinity and enhances the transcriptional activity of S186-phosphorylated BZLF1, which is required to induce the lytic phase of EBV's life cycle.

FIG. 1

Trans- and cis-acting elements of the lytic cycle of EBV. (A) (Upper part) The schematic modular structure of the functional domains in BZLF1 is similar to that of other members of the AP-1 family. BZLF1 consists of a transcriptional activation domain involved in DNA replication as well as in transcriptional activation, a basic DNA binding domain, and a dimerization domain which is responsible for homodimerization of BZLF1. (Lower part) Alignment of the amino acid sequences of the basic DNA binding domains of human c-Jun and BZLF1. Identical amino acids are indicated by vertical lines; double and single points mark similar and less similar residues, respectively. Arrowheads indicate amino acids in c-Jun that were shown to contact DNA bases (26). The serine at aa 186 in BZLF1 differs from the highly conserved alanine (open arrowhead) found at the equivalent position in the majority of bZip proteins. The amino acids serine 186, arginine 187, and lysine 188 in BZLF1, which form a conserved PKC motif, are highlighted. (B) Fine structure of oriLyt, with its cis-acting elements. oriLyt is located in a divergently transcribed promoter region which is shown schematically at the top. The flanking genes BHLF1 and BHRF1 are illustrated on the left and right (hatched boxes) together with their promoters (black bars). The open rectangles delineate the essential sequence elements which encompass the minimal oriLyt (59). The shaded regions represent poorly defined regions which function as auxiliary elements of oriLyt. An enlarged view of oriLyt is given at the bottom of the figure, including known functional sequence elements which are indicated by icons representing seven binding sites for BZLF1 (ZRE1 to ZRE7), one cluster of binding sites for the viral enhancer factor R, and the TATA and CCAAT boxes.

FIG. 2

TPA induces BZLF1-responsive gene expression. (A) BL41 cells were transiently transfected with the luciferase reporter plasmid pBHRF1-Luc (upper panel) or pBHLF1-Luc (lower panel). Where indicated, the BZLF1 expression plasmid pLPV-BZLF1-wt was cotransfected. The cells were treated with either TPA, the PKC inhibitor GF109203X, or a combination of both. Relative transcriptional activation was calculated on the basis of the luciferase activity in cells transfected with the reporter plasmid and vector (left column), which was set to one. (B) Neither TPA nor PKC inhibitor affect steady-state protein levels of BZLF1. Aliquots of the cell lysates which were analyzed in parallel for luciferase activity, as shown in panel A of this figure, were analyzed for the amount of BZLF1 by Western blotting. (C) Transcriptional activation of the mutant reporter construct p7xE2-BHRF1-Luc. This plasmid is identical to the reporter plasmid pBHRF-1-Luc shown in panel A of this figure, with the exception of the seven ZRE motifs which were replaced by seven E2 binding sites. Chimeric transcription factors [pCMV-c-Jun(trans):E2 and pCMV-BZLF1(trans):E2] which consist of the transactivation domains of c-Jun and BZLF1, respectively, fused to the DNA binding domain E2 of bovine papillomavirus were cotransfected with the reporter plasmid in comparison with pCMV-BZLF1-wt (which is unable to bind to the reporter plasmid) and a version of E2 itself (pCMV-E2-TR). Cells were treated subsequently with TPA or were left untreated, as indicated. Relative transcriptional activation is based on the measured luciferase activity in cells transfected only with the reporter plasmid (left column). TPA induction in cells transfected with the reporter plasmid alone led to an about twofold increase in luciferase activity.

FIG. 2

TPA induces BZLF1-responsive gene expression. (A) BL41 cells were transiently transfected with the luciferase reporter plasmid pBHRF1-Luc (upper panel) or pBHLF1-Luc (lower panel). Where indicated, the BZLF1 expression plasmid pLPV-BZLF1-wt was cotransfected. The cells were treated with either TPA, the PKC inhibitor GF109203X, or a combination of both. Relative transcriptional activation was calculated on the basis of the luciferase activity in cells transfected with the reporter plasmid and vector (left column), which was set to one. (B) Neither TPA nor PKC inhibitor affect steady-state protein levels of BZLF1. Aliquots of the cell lysates which were analyzed in parallel for luciferase activity, as shown in panel A of this figure, were analyzed for the amount of BZLF1 by Western blotting. (C) Transcriptional activation of the mutant reporter construct p7xE2-BHRF1-Luc. This plasmid is identical to the reporter plasmid pBHRF-1-Luc shown in panel A of this figure, with the exception of the seven ZRE motifs which were replaced by seven E2 binding sites. Chimeric transcription factors [pCMV-c-Jun(trans):E2 and pCMV-BZLF1(trans):E2] which consist of the transactivation domains of c-Jun and BZLF1, respectively, fused to the DNA binding domain E2 of bovine papillomavirus were cotransfected with the reporter plasmid in comparison with pCMV-BZLF1-wt (which is unable to bind to the reporter plasmid) and a version of E2 itself (pCMV-E2-TR). Cells were treated subsequently with TPA or were left untreated, as indicated. Relative transcriptional activation is based on the measured luciferase activity in cells transfected only with the reporter plasmid (left column). TPA induction in cells transfected with the reporter plasmid alone led to an about twofold increase in luciferase activity.

FIG. 3

A PKC inhibitor reduces the titer of EBV released from 293 cells. A fully recombinant EBV genome which carries the gene for GFP under control of the human cytomegalovirus immediate-early promoter/enhancer was established in a latent fashion in the epitheloid cell line 293, as described previously (15). Virus production was induced by transfecting pCMV-BZLF1-wt. Half of the transfected 293 cells were kept in the presence of the PKC inhibitor GF109203X or DMSO only, as a negative control, for 4 days. Cell-free supernatants were harvested from these 293 cells and 1/10 each of virus supernatants was used to infect Raji cells. The virus titers were determined by analyzing the number of green Raji cells by UV light microscopy, as shown in comparison to phase contrast micrographs. Supernatants which were obtained from 293 cells treated with PKC inhibitor show lower EBV titers (factor of five to six) than cells treated with DMSO only, as a negative control.

FIG. 4

BZLF1 is phosphorylated by PKCα at serine 186 in vitro. (A) Autoradiography of in vitro-labeled BZLF1. BZLF1-wt and point mutants BZLF1-T159A, BZLF1-S186A, and BZLF1-T159A/S186A were expressed as GST fusion proteins in E. coli and phosphorylated in vitro by PKCα. Following separation on an SDS gel, labeled proteins were analyzed by autoradiography. Only BZLF1-wt and the BZLF1 mutant T159A were efficiently phosphorylated by PKCα in vitro. (B) Coomassie stain of SDS gel from panel A of this figure. The amounts of the different GST-BZLF1 fusion proteins were comparable, as verified by Coomassie staining.

FIG. 5

Phosphopeptide mapping of BZLF1. (A) Bacterially expressed GST-BZLF1-wt was phosphorylated by PKCα in vitro prior to SDS-PAGE. The radiolabeled GST-BZLF1 band was identified by autoradiography, excised, and subjected to tryptic digestion. The released peptides were separated by two-dimensional electrophoresis and visualized by autoradiography. Two signals (x and y), which correspond to two partially digested peptides which carry S186, were detected. BZLF1-S186A failed to be phosphorylated by PKCα in vitro (Fig. 4) and consequently did not produce peptide spots (data not shown). (B) Two-dimensional tryptic phosphopeptide mapping of BZLF1-wt and BZLF1-S186A proteins from transiently transfected 293 cells which were metabolically labeled with [³²P]orthophosphate in vivo. The cells were left untreated or were incubated with TPA for 1 h prior to harvest. BZLF1 was immunoprecipitated and run on SDS-PAGE. The band was excised and subjected to tryptic digestion and two-dimensional phosphopeptide chromatography. BZLF1 is constitutively phosphorylated, as indicated by five spots labeled a through e which can be seen in all four panels. In addition, after TPA stimulation, BZLF1-wt but not BZLF1-S186A revealed two additional signals, labeled x and y. These signals correspond to the same spots seen with in vitro PKC-phosphorylated BZLF1, as shown in panel A of this figure. Arrowheads indicate the positions where the peptide samples were applied.

FIG. 6

Mutation of serine 186 reduces the activation of BZLF1-dependent promoters and their TPA inducibility. (A) (Upper panel) BL41 cells were transiently transfected with the luciferase reporter plasmid BHRF1-Luc and a constitutive expression vector for either BZLF1-wt, BZLF1-S186A, BZLF1-S186D, or BZLF1-S186E. After treatment with TPA for 12 h, the cells were harvested and lysed, and luciferase activity was determined and compared with extracts from untreated cells. The reporter activity is expressed as fold stimulation based on relative light units in cells transfected with reporter plasmid and vector (left column). BZLF1 mutants were expressed at levels similar to wild-type (wt) BZLF1, as confirmed by Western blotting (data not shown). (Lower panel) The same assay as described in the upper panel was repeated with 293 cells, yielding comparable results. (B) Transient transfection of BL41 cells with the minimal reporter construct 4xZRE5tk-Luc, containing four ZRE5 sites upstream of the basal thymidine kinase promoter. The cells were cotransfected with increasing amounts of pLPV-BZLF1-wt or pLPV-BZLF1-S186A with or without TPA treatment, and the luciferase activities of the different cellular extracts were determined as in panel A.

FIG. 7

Role of S186 mutation and TPA treatment in DNA binding affinity of BZLF1. (A) Gel retardation assay with bacterially expressed and purified BZLF1-wt and BZLF1-S186A. Equal concentrations of BZLF1 protein were incubated with labeled oligonucleotides with different ZRE or AP-1 motifs prior to electrophoresis. Each reaction sample was performed in duplicate with extracts from different preparations. For unclear reasons, complexes with the ZRE1 oligonucleotide were not detected. As a control, a GST fusion protein which lacks the DNA binding domain of BZLF1 (GST-BZLF1Δ167-245) was used. (B) Gel retardation assay with nuclear extracts of 293 cells transfected with BZLF1-wt and BZLF1-S186A and treated or not with TPA for 1 h. Nuclear extracts were incubated with labeled ZRE2/7 or ZRE5 oligonucleotides in the absence or presence of the BZLF1-specific antibody Z125. (C) Immunodetection of BZLF1 in nuclear extracts of the transfected 293 cells used in panel B demonstrates equal amounts of BZLF1 protein in the differently treated extracts.

FIG. 8

A putative cellular factor restores the DNA binding activity of BZLF1 phosphorylated in vitro. The relative binding affinity of bacterially expressed BZLF1 for different oligonucleotides with either AP-1 (left panel) or ZRE2/7 (right panel) motifs was analyzed after in vitro phosphorylation of BZLF1 by PKCα. The percentage of phosphate incorporation was about 50%, as determined with [³²P]ATP (data not shown). Comparable amounts of nonphosphorylated and phosphorylated BZLF1-wt were used in each case, as confirmed by Western blotting (data not shown). Nuclear extracts prepared from mock-transfected 293 cells were added prior to incubation with the appropriate oligonucleotide in the indicated samples. To demonstrate the presence of BZLF1 in the protein-DNA complexes, a BZLF1-specific antibody (Z125) was added to yield supershifted complexes. Numbers above and below the autoradiography signals mark the relative enhancement of signal intensities in corresponding sample pairs following incubation with mock-transfected nuclear extracts. Signal intensities were quantified by a phosphorimager.