Epstein-Barr Virus Late Gene Transcription Depends on the Assembly of a Virus-Specific Preinitiation Complex

ABSTRACT

During their productive cycle, herpesviruses exhibit a strictly regulated temporal cascade of gene expression that has three general stages: immediate early (IE), early (E), and late (L). Promoter complexity differs strikingly between IE/E genes and L genes. IE and E promoters contain cis-regulating sequences upstream of a TATA box, whereas L promoters comprise a unique cis element. In the case of the gammaherpesviruses, this element is usually a TATT motif found in the position where the consensus TATA box of eukaryotic promoters is typically found. Epstein-Barr virus (EBV) encodes a protein, called BcRF1, which has structural homology with the TATA-binding protein and interacts specifically with the TATT box. However, although necessary for the expression of the L genes, BcRF1 is not sufficient, suggesting that other viral proteins are also required. Here, we present the identification and characterization of a viral protein complex necessary and sufficient for the expression of the late viral genes. This viral complex is composed of five different proteins in addition to BcRF1 and interacts with cellular RNA polymerase II. During the viral productive cycle, this complex, which we call the vPIC (for viral preinitiation complex), works in concert with the viral DNA replication machinery to activate expression of the late viral genes. The EBV vPIC components have homologs in beta- and gammaherpesviruses but not in alphaherpesviruses. Our results not only reveal that beta- and gammaherpesviruses encode their own transcription preinitiation complex responsible for the expression of the late viral genes but also indicate the close evolutionary history of these viruses.

IMPORTANCE Control of late gene transcription in DNA viruses is a major unsolved question in virology. In eukaryotes, the first step in transcriptional activation is the formation of a permissive chromatin, which allows assembly of the preinitiation complex (PIC) at the core promoter. Fixation of the TATA box-binding protein (TBP) is a key rate-limiting step in this process. This study provides evidence that EBV encodes a complex composed of six proteins necessary for the expression of the late viral genes. This complex is formed around a viral TBP-like protein and interacts with cellular RNA polymerase II, suggesting that it is directly involved in the assembly of a virus-specific PIC (vPIC).

INTRODUCTION

Herpesviruses are enveloped viruses containing relatively large, double-stranded DNA genomes. They are divided into three subfamilies (alpha-, beta-, and gammaherpesviruses) according to sequence homology, cellular tropism, and productive cycle behavior under laboratory culture conditions. Nine herpesviruses have been identified in humans. Herpes simplex virus 1 (HSV-1) and 2 (HSV-2) and varicella-zoster virus (VZV) are alphaherpesviruses that due to neurotropism cause recurrent skin lesions, meningitis, and rare but very serious encephalitis in the case of HSV-1; human cytomegalovirus (HCMV), human herpesviruses 6A and 6B (HHV-6A and HHV-6B), and human herpesvirus 7 (HHV-7) are betaherpesviruses that cause severe diseases in patients with compromised immune function; and Epstein-Barr virus (EBV) and Kaposi's sarcoma-associated herpesvirus (KSHV) are gammaherpesviruses that are associated with a variety of cancers.

EBV is an orally transmitted herpesvirus that infects over 90% of the world's adult population. It is the causative agent of infectious mononucleosis and oral hairy leukoplakia, which results from EBV infection of epithelial cells along the side of the tongue (1). Moreover, EBV is also associated with several types of cancer, including Burkitt's lymphoma (BL), nasopharyngeal carcinoma (NPC), Hodgkin's disease (HD), T/NK cell lymphoma, and B cell lymphoproliferations and lymphoma in immunocompromised patients. Following infection, EBV persists for life in the host by establishing latency in memory B cells (2). In EBV-associated cancers, the virus is present in a latent form of viral infection; the viral genome is replicated once per cell cycle by the host cell machinery, and only a subset of the viral genes is expressed. In contrast, infection of normal oropharyngeal epithelial cells by EBV results in a productive or lytic infection that is similar to that observed in oral hairy leukoplakia lesions. EBV also periodically reactivates in B cells, particularly when they are stimulated by antigen, and B cells consequently differentiate into plasma cells (3). The switch from latent to productive infection is mediated by the EBV immediate early (IE) proteins EB1 (also called Zta or ZEBRA) and Rta (4–6). EB1 and Rta are transcription factors that cooperatively activate expression of the EBV early (E) genes, many products of which are involved in viral DNA replication (7). The virally encoded replication complex amplifies the viral genome via the lytic origin of replication (OriLyt) (8). Following viral DNA replication, the late (L) viral genes, encoding mainly proteins required for assembly, maturation, and release of infectious particles, are expressed (9). This coordinated cascade of viral gene expression during the viral productive cycle relies on the host RNA polymerase II (RNAP II) and is observed in all herpesviruses.

Whereas the regulation of immediate early and early gene expression has been extensively studied in herpesviruses, very little is known about the mechanisms regulating late gene expression. Immediate early and early gene promoters have the same structure as cellular gene promoters transcribed by RNAP II; they typically consist of distal regulatory sequences upstream of the TATA box which recruit TBP (TATA box-binding protein) at the origin of the formation of the preinitiation complex (PIC) on the promoters. The structure of the late gene promoters appears to be less complex, since it has been shown (in all cases studied so far) that a small proximal region is sufficient to control late gene expression (10–15). Interestingly, for the gammaherpesviruses, an unusual TATT sequence is often present where early genes and many cellular genes display a TATA box (11, 12, 15–17). In addition, while early genes are expressed independently of viral DNA synthesis, late genes are not expressed when viral DNA synthesis is blocked by specific inhibitors (10).

Until now, the mechanisms by which viral proteins are implicated in the regulation of the late gene promoters have not been precisely defined. In the case of HSV-1, several viral proteins (ICP4, ICP8, and ICP27) have been shown to be necessary for the efficient expression of late genes. These factors interact with the general transcription machinery and may thus function to facilitate the assembly of the transcription preinitiation complex on the late promoters. However, since these viral proteins also play an important role in the regulation of the early promoters, the specific molecular mechanisms by which they control the delayed expression of the late genes is unknown. For betaherpesviruses like CMV and gammaherpesviruses like mouse herpesvirus 68 (MHV-68), several viral proteins have been found to be involved in regulating late gene expression. These factors are expressed by beta- and gammaherpesviruses but not by the alphaherpesviruses. We have recently identified and characterized the function of one of these proteins: BcRF1 of EBV. BcRF1 was found, by in silico studies, to have structural homology with TBP from the thermophilic archaeon Pyrococcus woesei (17), and we recently showed that BcRF1 interacts specifically with the TATT sequence present on the promoters of late EBV genes, demonstrating for the first time that EBV encodes a TBP-like protein that is absolutely required for late viral gene expression. Moreover, we showed that BcRF1 alone had no effect on the expression of the late viral genes, suggesting that, to be functional, its coexpression of other viral proteins expressed during the viral productive cycle is required (18). These results, together with the finding that several HCMV or MHV-68 factors of unknown function are essential for late viral gene expression, suggest that control of late transcription via RNAP II in beta- and gammaherpesviruses relies on a specific late transcription preinitiation complex (19).

Here, we present the identification of the complete set of EBV genes required for late gene expression. We show that six viral proteins are necessary and sufficient for activation of the late viral genes. These six proteins, which include BcRF1, the viral TBP-like protein, form a complex that is required for EBV late gene transcription. This complex also recruits the cellular RNAP II enzyme. We have called this complex the viral preinitiation complex (vPIC), which is analogous to the cellular PIC. Moreover, we show that the HCMV proteins analogous to EBV vPIC factors are also sufficient to activate transcription from a model late gene promoter, although they are not able to complement an EBV vPIC from which individual proteins are missing. These data indicate that conserved beta- and gammaherpesvirus gene products control the transcription of late viral gene expression and suggest that the viral TBP-like proteins (BcRF1 in the case of EBV) recruit a virus-specific preinitiation complex on the TATT sequence in a manner similar to the recruitment of the cellular preinitiation complex by cellular TBP.

MATERIALS AND METHODS

Cell lines and transfections.

HEK293-EBV cells are derived from HEK293 cells infected with a bacterial artificial chromosome in EBV (BAC-EBV; kindly provided by W. Hammerschmidt). All cell lines were routinely grown in RPMI 1640 medium supplemented with 10% fetal calf serum. Transient transfections of the cells and derivatives were performed by electroporation (950 μF, 220 mV) using a Bio-Rad electroporator.

Recombinant plasmids.

An EB1 expression plasmid was used for initiation of the EBV productive cycle (20). The BcRF1 expression vector has previously been described (18). The entire BFRF2, BGLF3, BVLF1, BDLF4, and BDLF3.5 open reading frames (ORFs) were PCR amplified, cloned using the in-fusion system (Clontech) into the pCI expression plasmid (Promega) under the control of a CMV promoter, and fused at their 5′ ends to the Flag epitope. The entire HCMV UL49, UL79, UL87, UL91, UL92, and UL95 genomic sequences were PCR amplified with specific primers containing unique restriction sites, and the FLAG epitope was fused to their 5′ ends. PCR products were digested with the appropriate restriction enzymes and cloned into the corresponding sites of the pCIneo vector (Promega) under the control of a CMV promoter. The primer sequences are given in Table 1. The pTATT_BcLF1-Luc reporter plasmid has previously been described (18). It contains sequences from position −38 to −4 from the viral late gene BcLF1, including the TATT element.

TABLE 1.

List of the different primers used in this study

Primer name	Sequence
BFRF2 ORFs	AAGGGATCCCCTCGAGAAATGGCGTTATTCTTGGCGCG
BFRF2 ORFas	CCGGGTCGACTCTAGATTAGGAAGCAGGGGACTGTC
BGLF3 ORFs	AAGGGATCCCCTCGAGAAATGTTCAACGCGGTC
BGLF3 ORFas	CCGGGTCGACTCTAGACTACTCATCTTCATAAGTC
BVLF1 ORFs	AAGGGATCCCCTCGAGAAATGTTAATGGGACTGG
BVLF1 ORFas	CCGGGTCGACTCTAGACTACCTGGCCTCCCCGGTG
BDLF4 ORFs	AAGGGATCCCCTCGAGAAATGTCAGACCAAGGCCG
BDLF4 ORFas	CCGGGTCGACTCTAGATCAACACTTGGTTGTCAATGTGG
BDLF3.5 ORFs	AAGGGATCCCCTCGAGAAATGTCTGCCCCCGGATGCTC
BDLF3.5 ORFas	CCGGGTCGACTCTAGATCAATCGGCCTTGGTCTGAC
VirusΔBFRF2sa	CAGGCCCTTGGGGAGCGGGGGTTCTCCAGGCTCCTGGATCTGGGGCTGGCGAAAAGTGCCACCTGCAGA
VirusΔBFRF2asa	GGAATTGAGCCTGCCTCCTACACCTGTCTGCTTTCCAAAGTTCAAGCAGTCAGGAACACTTAACGGCTG
BFRF2 a	ATGGCGTTATTCTTGGCGCG
BFRF2 b	CTCACCGGATTCAGTCG
BFRF2 c	TAGATTGTCGCACCTG
BFRF2 d	ATTAGGAAGCAGGGGACTGTC
BFRF2 e	GCCTCGTGAAGAGGTTCTTG
BMRF1s qPCR	AGGAGTGCTGCAGGTAAACC
BMRF1as qPCR	GCTCTGGTGATTCTGCCACT
BDLF1s qPCR	CGCAGACATGCTCGATGTA
BDLF1as qPCR	GTAGTGGTGCCCCAGGTATG
BcLF1s qPCR	GGCTCAGTCTAAGG
BcLF1as qPCR	AGGTGGGCTGACACAGACTT
GAPDHs qPCR	AGCCACATCGCTCAGACAC
GAPDHas qPCR	GCCCAATACGACCAAATCC
BFRF3s qPCR	AGAGGCAGAGAGCCAGTGTG
BFRF3as qPCR	CCGGAGGCTGCTAATAGATG
FlUL49-S	CGCCGCCACCATGGACTACAAAGACGATGACGATAAGGCCAGTCGTCGTCTCCGAC
UL49-AS	AAAAGCGGCCGCGGTGTTAGACATGGGG
FlUL79S	AAAAACGCGTGCCGCCACCATGGACTACAAAGACGATGACGATAAGATGATGGCCCGCGACG
UL79AS	ATGAGCGGCCGCGCTGTAC
FlUL87S	AAAAGAATTCGCCGCCACCATGGACTACAAAGACGATGACGATAAGGCCGGCGCTGCGC
UL87AS	AAAAGTCGACGCTCCTCCGGACGAAAC
FlUL91S	TGCCGCCACCATGGACTACAAAGACGATGACGATAAGAACTCGTTGCTGGCGGAAC
UL91AS	AAAAGCGGCCGCGTTCTGGACGTGCC
FlUL92-S	CGCCGCCACCATGGACTACAAAGACGATGACGATAAGTGCGACGCCTCGGG
UL92-AS	AAAAGCGGCCGCGCTTCAAACGCC
FlUL95S	AAAAGCTAGCCGCCACCATGGACTACAAAGACGATGACGATAAGATGATGGCGGCGGCG
UL95AS	AAAAACGCGTATCACGTCCTTTAAGAGCTGTTTGTTG
IGUSB/F29	GTGCTGGGGAATAAAAAGGGG
IGUSB/F29	ACTCGGGGAGGAAGGGACAC

^aFor the ΔBFRF2 virus primers, the underlined sequences allowed PCR-mediated amplification of the kanamycin resistance gene. The other primer sequences are homologous to BFRF2 sequences and are used for recombination of the amplified PCR product within the BFRF2 coding sequence.

Generation of a BFRF2-KO recombinant EBV.

To generate a BFRF2 knockout mutant in the context of the EBV genome, we used the maxi-EBV technology described in the work of Delecluse et al. (21). The EBV_ΔBFRF2 mutant was constructed by replacing part of the BFRF2 ORF (B95.8 coordinates 47668 to 48700; GenBank accession number AIE88895) with the kanamycin (kan) resistance gene using homologous recombination (22). To remove the kanamycin cassette, the Escherichia coli bacteria harboring BAC-EBV_ΔBFRF2 were transformed with the plasmid pCP20, encoding red recombinase, which allowed the cassette's elimination. Recombinant EBV plasmid DNA (10 μg) was then transfected into HEK293 cells using polyethylenimine (PEI) reagent. One day posttransfection, hygromycin (100 μg/ml) was added to the culture medium for selection of stable HEK293 clones carrying the EBV recombinant plasmid. Outgrowing green fluorescent protein (GFP)-positive colonies were amplified for further investigation. The cell clone used in this study is referred to as HEK293_EBVΔBFRF2.

Induction of the virus productive cycle.

Producer cell clones HEK293_EBV (kindly provided by W. Hammerschmidt) and HEK293_EBVΔBFRF2 were transfected with an EB1 expression plasmid (1 μg/plate) to induce the viral productive cycle. In trans-complementation assays, HEK293_EBVΔBFRF2 cells were cotransfected with 1 μg of plasmid pCI-Flag-BFRF2 or the empty plasmid. Two days posttransfection, cells were harvested for analysis of viral gene expression.

Viral DNA replication analysis.

EBV-infected cells were collected 72 h posttransfection, resuspended in lysis buffer (23), and incubated at 4°C overnight. DNA was extracted with phenol-chloroform, treated with RNase A (100 μg/ml), precipitated with ethanol, and then resuspended in nuclease-free water. Viral DNA levels were determined by quantitative PCR (qPCR).

Western blot analysis.

Antibodies used for Western blotting were an anti-BZLF1 monoclonal antibody (MAb), Z125 (20), a rabbit polyclonal serum directed against the EB2 protein (24), an anti-Flag M2 MAb (Sigma; F3165), an anti-Flag polyclonal antibody (Sigma; F7425), and an anti-RNAP II polyclonal antibody (Santa Cruz; N-20 sc-899), as well as anti-mouse or anti-rabbit horseradish peroxidase-conjugated secondary antibodies (GE Healthcare Life Sciences). Immune complexes were visualized using enhanced-chemiluminescence (ECL) reagent (Pierce).

RNA extraction and RT-qPCR analysis.

Total RNA was extracted and purified using the NucleoSpin RNA kit (Macherey-Nagel) and reverse transcribed using qScript cDNA SuperMix (Quanta Biosciences). Real-time reverse transcription-PCR (RT-qPCR) was performed using FastStart universal SYBR green master mix (Rox; Roche Molecular Biochemicals) and primers listed in Table 1. Cellular glyceraldehyde-3-phosphate dehydrogenase (GAPDH) RNA was quantified as an internal control for the amounts of RNA. Quantification of the RT-qPCR results was done as previously described (25).

Luc assays.

Firefly luciferase (Luc) activity from transfected cells was measured in a Veritas luminometer (Turner Biosystems) using the Firefly luciferase assay system (Promega). Luciferase activity was measured for identical amounts of total protein as evaluated by the Bradford assay.

In vivo GST pulldown assays.

The pCi-GST.Flag or pCi-GST.Flag.BcRF1 expression plasmid was transfected into HEK293 cells together, or not, with expression vectors for the vPIC factors. Forty-eight hours after transfection, cells were lysed in HNTG buffer (20 mM HEPES, pH 7.9, 180 mM NaCl, 10% glycerol, 0.1 mM EDTA, 0.1% NP-40, 1 mM dithiothreitol [DTT], 1 mM phenylmethylsulfonyl fluoride [PMSF]). Glutathione-Sepharose beads saturated with bovine serum albumin (BSA) were added to the lysate in order to purify the glutathione S-transferase (GST) fusion proteins and their associated proteins. After 4 h of incubation, the beads were washed 5-fold with HNTG buffer and bound proteins were fractionated by SDS-PAGE and analyzed by Western blotting. The vPIC proteins were visualized with anti-Flag antibodies, and the RNAP II protein was visualized with rabbit polyclonal antibody sc-899 (Santa Cruz).

RESULTS

Identification of viral proteins required for late gene expression.

We have previously shown that BcRF1 is necessary, but not sufficient alone, to activate late gene transcription (18), suggesting that other viral proteins are required. In the case of MHV-68, five proteins (ORF18, ORF24, ORF30, ORF31, and ORF34) were found to be necessary for the expression of the MHV-68 late genes. Analogs for these proteins in CMV (UL79, UL87, UL91, UL92, and UL95, respectively) were also found to be necessary for late gene expression (26–33). Interestingly, EBV also encodes five putative analogs (BVLF1.5, BcRF1, BDLF3.5, BDLF4, and BGLF3, respectively). We have already shown by the generation of a recombinant virus (EBV_ΔBcRF1) that BcRF1 is required for late gene expression (18). Together, these data suggest that a conserved set of proteins implicated in the control of late gene expression is conserved between beta- and gammaherpesviruses. Interestingly, these genes are not conserved in the alphaherpesviruses. Based on this observation, we established a list of the early viral proteins that are conserved in the betaherpesvirus (CMV and HHV6) and gammaherpesvirus (EBV, KSHV, and MHV-68) subfamilies but absent in HSV-1 (Table 2). From this comparison, we identified two proteins in addition to those cited above, called BTRF1 and BFRF2 in the case of EBV. The analog of BTRF1 in MHV-68 (ORF23) has previously been shown to be the product of a late gene and dispensable for virus production both in cell culture and in vivo (34). Interestingly, however, BFRF2 is an uncharacterized protein of unknown function. We thus decided to further explore the role of this protein in the expression of the late viral genes of EBV.

TABLE 2.

List of analogous genes between beta- and gammaherpesviruses which are not present in alphaherpesviruses^a

EBV strain	Analogous gene in:				Reference(s)
	KSHV	MHV-68	CMV	HHV-6
BVLF1	ORF18	ORF18	UL79	U52	33, 26
BcRF1	ORF24	ORF24	UL87	U58	18, 28
BDLF3.5	ORF30	ORF30	UL91	U62	30, 27
BDLF4	ORF31	ORF31	UL92	U63	31, 29
BGLF3	ORF34	ORF34	UL95	U67	27
BFRF2	ORF66	ORF66	UL49	U33	This publication
BTRF1	ORF23	ORF23	UL88	U59	34

^aFor more information, see reference 45. References for the genes' implications in late viral gene expression are indicated in the extreme-right column.

Construction of a BAC-EBV recombinant clone carrying a deletion in the BFRF2 gene.

In order to study the function of BFRF2 in the EBV productive cycle, a BAC-EBV recombinant named BAC-EBV_ΔBFRF2 was constructed in a manner that did not interfere with the expression or functions of adjacent genes. For this, we introduced a large deletion in the BFRF2 ORF (Fig. 1A). The BFRF2 early gene is located between the early BFRF1 gene and the late BFRF3 gene. The BFRF2 gene is transcribed from the same promoter as the BFRF1 gene, and the ORF of the BFRF2 gene has been sized to 1,775 bp, coding for a 64-kDa protein. To generate the BAC-EBV_ΔBFRF2 mutant, a 1,032-bp fragment from the BFRF2 gene (B95.8 coordinates 47668 to 48700) was exchanged for the kanamycin resistance gene sequence by homologous recombination in E. coli while leaving the first 148 bp from the BFRF2 ORF intact in order to avoid disruption of the transcriptional regulation of its neighboring genes. In a second step, the kanamycin resistance gene was deleted to generate the BAC-EBV_{ΔBFRF2ΔKana} recombinant (Fig. 1A). The genome integrity of all mutants and control wild-type BAC-EBV was verified by BamHI restriction enzyme digestion. As expected from the cloning strategy, the restriction profile confirmed that the 7,396-bp BamHI fragment that contains the BFRF2 gene had been replaced by a 7,442-bp fragment in the ΔBFRF2 mutant and reduced to a 6,365-bp fragment after deletion of the kanamycin gene (Fig. 1B). Moreover, PCR amplification of the BFRF2 region with primers located on each side of the BFRF2 ORF gave a fragment of 1,821 bp for BAC-EBV_ΔBFRF2 and 744 bp for BAC-EBV_{ΔBFRF2ΔKana}, compared with the 1,780-bp fragment coming from wild-type BAC-EBV (Fig. 1C, primers a and d). As expected, PCR amplification with primers inside the kanamycin gene gave a fragment only with the BAC-EBV_ΔBFRF2 recombinant (1,288 bp with primers c and d or 591 bp with primers a and b). BFRF2 ORF internal primers amplified only the wild-type BAC-EBV genome, giving a fragment of 230 bp, as expected (Fig. 1C, primers e and f). BAC-EBV_{ΔBFRF2ΔKana} DNA was prepared and transfected into HEK293 cells in order to establish an HEK293 cell line harboring this mutated virus, hereinafter referred to as HEK293_EBVΔBFRF2. We screened several clones for their permissivity for the EBV productive cycle. The cell clones were transfected with the EB1 expression plasmids in order to induce the viral productive cycle, and after 48-h cell extracts were prepared, we verified by Western blotting expression of the viral early protein EB2 (the product of the BMLF1 gene). EB2 was used as a marker for the induction of the viral productive cycle. We then selected the cell clone with the highest expression level of EB2 to perform further experiments.

FIG 1.

Construction of the BacEBV_ΔBFRF2 recombinant. (A) Schematic representation of the EBV region surrounding the BFRF2 gene (1). Homologous recombination between the wild-type EBV genome (p2089) and a PCR fragment carrying the kanamycin gene flanked by 50 bp of BcRF1 sequences was performed in the DH10B E. coli strain containing the pKD46 plasmid. Clones carrying the recombinant EBV were selected for chloramphenicol and kanamycin resistance. The resulting BAC-EBV_ΔBFRF2 recombinant had the kanamycin gene inserted between positions 47668 and 48700 in the BFRF2 gene (2). The kanamycin gene was then deleted by homologous recombination using the pCP20 vector in order to generate the BAC-EBV_{ΔBFRF2ΔKana} recombinant used in this study (3). (B) Restriction enzyme analysis of the BAC-EBV_ΔBFRF2 plasmid DNA. The BamHI restriction pattern of BAC-EBV_ΔBFRF2 (lane 2) and BAC-EBV_{ΔBFRF2ΔKana} (lane 3) mutant DNA was compared to that of wild-type BAC-EBV DNA (lane 3). An asterisk indicates the modified DNA pattern. Lane M, molecular markers. (C) PCR amplification analysis of the BAC-EBV_ΔBFRF2 (lanes 2) and BAC-EBV_{ΔBFRF2ΔKana} (lanes 3) (ΔF2) clones compared to the BAC-EBV clone (lane 1) carrying the wild-type EBV genome. PCRs were performed using primers indicated by small arrowheads in panel A (a, b, c, d, e, and f) and listed in Table 1.

BFRF2 is required during the late phase of the viral productive cycle.

In order to test the production of progeny virus by HEK293_EBVΔBFRF2 cells, they were transfected with an expression plasmid encoding the EBV transcription factor EB1 to induce the viral productive cycle, either alone or together with an expression plasmid for the BFRF2 protein. Two days after transfection, cell supernatants were used to infect Raji cells. The presence of infectious EBV was visualized indirectly by GFP expression in infected Raji cells as previously described (21, 35). As shown in Fig. 2A, 2 days postinfection, no green Raji cells were detected with supernatants from cells transfected with the EB1 expression vector alone. However, with the supernatants from cells transfected with expression plasmids for both EB1 and BFRF2, a large number of green Raji cells was observed (Fig. 2A), indicating that ectopic expression of BFRF2 can rescue virus production from the HEK293_EBVΔBFRF2 cells. Quantification by fluorescence-activated cell sorting (FACS) analysis showed that an average of 8% of the Raji cells were GFP positive after infection with the supernatant of HEK293_EBVΔBFRF2 cells transfected with expression plasmids encoding EB1 and BFRF2, whereas no GFP-positive Raji cells were detected after infection with the supernatant of HEK293_EBVΔBFRF2 cells transfected with expression plasmids encoding EB1 alone. These results clearly demonstrate that the EBV BFRF2 gene is essential for the production of infectious EBV particles.

FIG 2.

BFRF2 is necessary for virion production. (A) Raji cells were incubated with supernatants from HEK293_EBVΔBFRF2 cells transfected with an expression plasmid encoding the EB1 protein in order to induce the productive cycle together (or not) with an expression plasmid encoding the Flag.BFRF2 protein to complement BFRF2 deficiency. (Left side) The GFP fluorescence of Raji cells was analyzed 48 h after infection under UV light. (A, sub panels a and b) Phase contrast; (A, sub panels c and d) UV light. (Right side) Immunoblot analysis of the EB2, Flag.BFRF2, and EB1 proteins expressed in HEK293_EBVΔBFRF2 cells that either had not been transfected (lane 1), had been transfected with an EB1 expression plasmid (lane 2), or had been cotransfected with expression plasmids for both EB1 and BFRF2 (lane 3). Western blots were revealed using a polyclonal anti-Flag antibody, an anti-EB2 polyclonal rabbit antibody, or an anti-EB1 MAb. * indicates an unspecified protein recognized by the anti-EB2 serum. (B) BFRF2 is not required for viral DNA replication. Viral DNA in HEK293_EBV and HEK293_EBVΔBFRF2 cells transfected with EB1 and Flag.BFRF2 expression plasmids, as indicated in the figure, was quantitated by qPCR. The amount of viral DNA in the induced cells (transfected with the EB1 expression plasmid) is expressed relative to the amount of viral DNA present in the uninduced cells. (C, top) BFRF2 enhances expression of the late viral genes. Total RNA from HEK293_EBV (wild type [WT]) or HEK293_EBVΔBFRF2 (ΔF2) cells that either had not been transfected (−), had been transfected with an EB1 expression plasmid (EB1), or had been cotransfected with expression plasmids for EB1 and Flag.BFRF2 (EB1 + BFRF2) was purified, reverse transcribed, and analyzed by qPCR using specific primers for either the BMRF1 early gene or the BDLF1 and BcLF1 late genes. (Bottom) Western blot of HEK293_EBV or HEK293_EBVΔBFRF2 cells either not transfected (−) or transfected with an EB1 expression plasmid or with expression plasmids for both EB1 and Flag.BFRF2. Protein extracts were analyzed by Western blotting using a polyclonal anti-Flag antibody and an anti-EB1 MAb. The asterisk indicates an unspecified protein detected by the anti-Flag antibody.

We hypothesized that, as a putative late gene expression regulator, BFRF2 should affect neither the expression of the viral early genes nor the OriLyt-dependent replication of the viral genome. In order to test this, HEK293_EBVΔBFRF2 cells or HEK293_EBV cells as a control were either mock transfected or transfected with an expression vector for EB1 so as to activate EBV's productive cycle, together, or not, with a BFRF2 expression plasmid. Forty-eight hours posttransfection, expression of the viral early protein EB2 (BMLF1 or Mta) was monitored by Western blotting (Fig. 2A) and viral DNA was purified and quantified by qPCR (Fig. 2B). As expected, the viral early protein EB2 was expressed in cells induced in the productive cycle by ectopic expression of the viral immediate early protein EB1, and its expression was not significantly modified when BFRF2 was coexpressed in HEK293_EBVΔBFRF2 cells (Fig. 2A). Quantification of the viral DNA showed that induction of the productive cycle by EB1 led to a strong increase in the amount of viral DNA present in HEK293_EBVΔBFRF2 cells, comparable to that observed in HEK293_EBV cells. However, coexpression of BFRF2 in HEK293_EBV or HEK293_EBVΔBFRF2 cells did not change the amount of newly replicated viral DNA, demonstrating that the BFRF2 protein is not required for OriLyt-dependent viral DNA replication (Fig. 2B). Taken together, these results demonstrate that despite being an early protein, BFRF2 neither affects expression of the early viral proteins nor participates in viral DNA replication.

Next, in order to define BFRF2's impact on viral late gene transcript levels, we analyzed the accumulation of viral RNA for some early and late genes. Total RNA was extracted from HEK293_EBV or HEK293_EBVΔBFRF2 cells that were mock transfected, transfected with an EB1 expression vector, or transfected with both an EB1 and a BFRF2 expression vector. RNAs were then quantified by RT-qPCR using primers specific for the early BMRF1 gene or the late BDLF1 and BcLF1 genes. In mock-transfected HEK293_EBV or HEK293_EBVΔBFRF2 cells, the early or late mRNAs were, as expected, not detectable, whereas in cells transfected with an EB1 expression plasmid, the early BMRF1 transcript accumulated at comparable levels for the mutant and control viruses (Fig. 2C). In contrast, the late BDLF1 and BcLF1 mRNA levels were remarkably low in cells harboring the BFRF2-mutated virus compared to those in cells infected with the wild-type BAC-EBV. However, after coexpression of BFRF2 in HEK293_EBVΔBFRF2 cells, expression of the late BDLF1 and BcLF1 mRNAs was clearly enhanced (Fig. 2C).

The implication of BFRF2 in late viral gene expression was then confirmed in a transient-transfection assay by using the reporter vector pTATT_BcLF1-Luc, which carries typical late promoter sequences (Fig. 3A). The results (Fig. 3B) showed that in HEK293_EBVΔBFRF2 cells, induction of the pTATT_BcLF1-Luc reporter required both the activation of the viral productive cycle by EB1 and the expression of the BFRF2 protein. These results are very similar to those obtained using HEK293_EBVΔBcRF1 cells, where induction of the pTATT_BcLF1-Luc reporter required both activation of the viral productive cycle by EB1 and expression of the BcRF1 protein (Fig. 3B) (18). Taken together, the above-described results demonstrate that BFRF2, like BcRF1, is necessary but not sufficient for efficient transcription of the late viral genes.

FIG 3.

BFRF2 is required for late gene expression. (A) Schematic representation of the luciferase reporter vector used (pTATT_BcLF1-Luc). (B) HEK293_EBV, HEK293_EBVΔBcRF1, or HEK293_EBVΔBFRF2 cells were either transfected with the reporter vector alone (−) or cotransfected with an expression plasmid for EB1, Flag.BcRF1, or Flag.BFRF2, as indicated. The amount of luciferase expressed was measured 24 h after transfection. The relative luciferase fold activation was calculated relative to the value obtained with the pTATT_BcLF1-Luc reporter transfected alone. The experiment was performed 3 times, and the error bars represent standard deviations. Protein extracts from the experiment whose results are shown in panel B were tested for EB1, BcRF1, and BFRF2 expression by immunoblotting using an anti-EB1 MAb, an anti-Flag MAb (BcRF1), and a polyclonal anti-Flag antibody (BFRF2).

Six viral proteins are necessary and sufficient for the activation of the late viral genes of EBV.

The data previously obtained with different beta- and gammaherpesviruses and the above-described results suggest that six viral proteins are required for the activation of the late viral genes of EBV. In order to show that the products of these six viral genes are not only necessary but also sufficient for the transcriptional activation of the late viral genes, we transfected their corresponding expression vectors into EBV-negative HEK293 cells, together with the pTATT_BcLF1-Luc reporter plasmid. The results of the luciferase assays (Fig. 4A) show that none of these viral proteins alone are able to activate the expression of the luciferase reporter. However, when the cells were cotransfected with the six different expression vectors, an average 10-fold induction of luciferase activity was observed, suggesting that, together, the six proteins can activate late viral gene expression. When the cells were cotransfected with various combinations of only five of the expression vectors, the luciferase activity was not induced (Fig. 4C). All six viral proteins are thus required for activation of viral late gene expression. In contrast, coexpression of the six viral proteins had no effect on the level of luciferase expressed under the control of the viral early pBMRF1 promoter, whereas EB1 strongly induced this expression (Fig. 4E). However, EB1 was unable to induce the expression of the luciferase gene controlled by the BcLF1 minimal promoter (Fig. 4A). Taken together, these results suggest that the six viral proteins are required for the activation of the late viral genes of EBV.

FIG 4.

Six viral proteins are required for activation of the late viral genes. (A) The luciferase (Luc) reporter vector pTATT_BcLF1-Luc was transfected into the EBV-negative cell line HEK293, together with expression vectors for the six different proteins, separately or combined. An EB1 expression vector was also used in order to show that the response of the TATT_BcLF1 promoter was specific for the other six proteins. (B) Proteins expressed in HEK293 cells transfected with the different expression plasmids encoding the Flag-tagged proteins used in panel A were analyzed by Western blotting using a monoclonal anti-Flag antibody. White asterisks label the expected positions of the different proteins on the gel. − indicates the missing protein. (C) HEK293 cells were transfected, together with the pTATT_BcLF1-Luc reporter, with expression vectors for the six different viral proteins combined (all) or with only five expression vectors as indicated (“all-” indicates the missing expression vector). (D) Western blot corresponding to HEK293 cells transfected with expression vectors for all 6 proteins or with different combinations of five of the proteins. All the proteins were tagged with a Flag epitope, and the blot was incubated with an anti-Flag MAb. Black asterisks label the positions of proteins destabilized in the absence of one component (the monoclonal anti-Flag antibody does not react with Flag-BDLF3.5). (E) The pBMRF1-Luc reporter was transfected into the EBV-negative cell line HEK293, together with combined expression vectors for the six different proteins (all) or an EB1 expression vector. (A, C, and E) The amount of luciferase expressed was measured 24 h after transfection. The relative luciferase fold activation was calculated relative to the value obtained with the pTATT_BcLF1-Luc or the pBMRF1-Luc reporter transfected alone. The experiment was performed 3 times, and the error bars represent standard deviations. (F) BGLF3 and BVLF1 are stabilized by coexpression of BFRF2 and BDLF3.5, respectively. HEK293 cells were transfected with a Flag.BGLF3 expression vector either alone or in combination with a Flag.BFRF2 expression vector or with a Flag-BVLF1 expression vector either alone or in combination with a Flag-BDLF3.5 expression vector. The immunoblot was revealed with a polyclonal anti-Flag antibody. (G) Total RNA from the transfected cells was extracted, and RT-PCR was performed in order to show that expression of BFRF2 and BDLF3.5 did not increase the amount of BGLF3 and BVLF1 mRNA, respectively. Primers specific for the cellular GusB mRNA were used to control for the absence of contaminating DNA in the assay and to ensure that the same amount of mRNA was used for the RT.

Further analysis of the proteins expressed from cells transfected with the different expression vectors separately showed that most of the proteins expressed in the assay are individually unstable. Except for BcRF1 and BDLF4, all the other proteins were barely detectable when expressed alone. Furthermore, although detectable, BcRF1 was very sensitive to degradation (Fig. 4B). However, when coexpressed, all the proteins, except BDLF3.5, likely due to its small size, were easily detectable (Fig. 4B, lane 8), suggesting that these proteins form a complex which allows their stabilization. Interestingly, when HEK293 cells were transfected with various combinations of five of the viral proteins, we found that the expression of BFRF2 was required to observe detectable levels of BGLF3 (but not the reverse) and that BVLF1 was stabilized by the coexpression of BDLF3.5 (Fig. 4D and F). It is important to note that BFRF2 and BDLF3.5 do not act at the transcriptional level on the accumulation of, respectively, BGLF3 and BVLF1, as shown by RT-PCR analysis (Fig. 4G). Taken together, these results suggest that interactions between the different proteins are necessary for their stabilization and also that these six proteins are required for the formation of a basic transcription initiation complex able to activate the expression of the late viral genes. Because of the presence in the complex of BcRF1, the viral TBP-like protein, and by analogy to the cellular preinitiation complex formed around the cellular TBP, we decided to call this complex vPIC, for viral preinitiation complex.

The six viral proteins of the vPIC form a complex associated with RNAP II.

In order to further analyze the interactions between the vPIC components, we transfected HEK293 cells with a GST-Flag or a GST-Flag.BcRF1 expression plasmid together with expression vectors for the vPIC viral components (each tagged with the Flag epitope). Forty-eight hours after transfection, the cell lysates were used in a GST pulldown assay. Five of the six vPIC components were found associated with GST-Flag.BcRF1 but not with GST (Fig. 5A). However, it was not possible to conclude this for the sixth protein, BDLF3.5, as this protein was already barely detectable in the input. It is interesting to note that BcRF1 itself was pulled down with GST-Flag.BcRF1, suggesting that BcRF1 either forms homodimers or is engaged in interactions with several components of the complex. This result confirms that the six viral proteins are involved in the same complex. This was also seen in in vitro GST pulldown assays in which in vitro-translated proteins labeled with [³⁵S]methionine were incubated with the different vPIC proteins fused to GST and produced in bacteria (Fig. 5B). The results of these in vitro GST pulldown assays show that each protein appears to interact both with itself and with all the others, although interactions detected with BVLF1 were much weaker. The fact that BVLF1 was very efficiently pulled down with GST-Flag-BcRF1, together with the rest of the vPIC components from HEK293 cells (Fig. 5A), suggests that BVLF1 has concomitant interactions with several vPIC components.

FIG 5.

GST pulldown of the vPIC. Extracts from HEK293 cells transfected with a GST-Flag (GST) or a GST-Flag.BcRF1 (GST-BcRF1) expression plasmid together with expression vectors for components of the vPIC (all tagged with the Flag epitope) were incubated with glutathione-Sepharose beads. The purified proteins were analyzed by immunoblotting with a mix of monoclonal and polyclonal anti-Flag antibodies in order to identify the components of the vPIC associated with GST-Flag.BcRF1 (A) or with an anti-RNAP II antibody in order to analyze the interaction between BcRF1 and the RNAP II complex (C). GST-Flag and GST-Flag.BcRF1, which are very weakly expressed in the cells, are seen only after GST pulldown. Dark-gray arrows indicate their positions. Again, Flag.BDLF3.5 was not efficiently detected. II0 indicates the position of the hyperphosphorylated form of the RNAP II enzyme and IIA its unphosphorylated form. (B) Summary of the interactions found between the different components of the vPIC as determined by in vitro GST pulldown assay. In vitro-translated ³⁵S-labeled BcRF1, BGLF3, BVLF1, BDLF4, and BFRF2 proteins were incubated with purified GST, GST-BcRF1, GST-BGLF3, GST-BDLF4, GST-BFRF2, or GST-BDLF3.5 bound to glutathione-Sepharose beads. The bound proteins were analyzed by SDS-PAGE and visualized by autoradiography. None of the in vitro-translated ³⁵S-labeled proteins interacted with GST.

In order to analyze the interaction between the vPIC and the RNA polymerase II machinery, proteins eluted from a GST-Flag.BcRF1 pulldown assay, performed with transfected HEK293 cells lysates, were analyzed by Western blotting using a specific anti-RNAP II antibody. In these experiments (Fig. 5C), we found that the cellular RNAP II large subunit was specifically pulled down with the GST-Flag.BcRF1 protein (lane 3) but not with GST (lane 1). Interestingly, the interaction appeared to be more important when the full vPIC was expressed in the cells (lane 4). Moreover, the hyperphosphorylated II0 form of RNAP II (the active form) was the only RNAP II form found associated with GST-Flag.BcRF1.

Taken together, these results demonstrate that the vPIC components are involved in a complex network of mutual interactions and, most importantly, that this viral complex associates with the active form of RNAP II.

Viral DNA replication is required for expression of the endogenous late viral genes.

As reported above, in transient-transfection assays, the vPIC can induce the expression of a reporter gene under the control of a late viral promoter. However, in the context of the whole EBV genome, it is well known that viral DNA replication is a prerequisite for the expression of the late viral genes. Accordingly, when HEK293_EBV or Raji cells were transfected with the vPIC alone, the late BDLF1 and BFRF3 viral mRNAs were not expressed (Fig. 6A) despite the vPIC being functional, as could be demonstrated by cotransfection with the pTATT_BcLF1-Luc reporter (Fig. 6B). As expected, expression of the early BMRF1 gene was also not induced by the vPIC. In contrast, when HEK293_EBV cells were transfected with an EB1 expression vector in order to induce the viral productive cycle, both the early BMRF1 gene and the late BDLF1 and BFRF3 genes were expressed. However, in the case of Raji cells, which are unable to replicate the viral genome, only the early BMRF1 gene was efficiently expressed. The late BDLF1 and BFRF3 genes were not expressed, although components of the vPIC are produced as part of the early genes. From these results, we conclude that vPIC expression is not sufficient for the induction of the late viral genes from the EBV genome. Another event, probably replication of the viral DNA, is required for the vPIC to be fully functional on the endogenous late genes.

FIG 6.

The vPIC was unable to induce late gene expression from the endogenous viral genome. (A) Total RNA from HEK293_EBV or Raji cells either mock transfected (−) or transfected with the EB1 expression vector or the six expression plasmids encoding the vPIC factors was quantified by RT-qPCR using primers specific for either the BMRF1 early gene or the BDLF1 and BFRF3 late genes. The experiment was performed 3 times, and the error bars represent standard deviations. (B) EB1 and vPIC induce late gene expression in a transient-transfection assay of EBV-positive cells. HEK293_EBV or Raji cells were transfected with the pTATT_BcLF1-Luc reporter plasmid either alone (−) or together with the EB1 expression vector or the six expression plasmids encoding the vPIC components. The experiment was performed 3 times, and the error bars represent standard deviations. Protein extracts from the experiment whose results are shown in panel B were tested for expression of EB1, and the vPIC components were tested by immunoblotting using an anti-EB1 MAb or a mix of monoclonal and polyclonal anti-Flag antibodies (for the vPIC proteins). A star indicates an unspecified band recognized by the anti-Flag antibody.

This was confirmed by the finding that treatment of HEK293_EBV cells in which the viral productive cycle had been induced with phosphonoacetic acid (PAA), an inhibitor of EBV DNA replication, completely inhibited the expression of the late viral genes even when the vPIC was ectopically expressed (Fig. 7B and C). However, it had only a limited effect (50% decrease) on the level of luciferase expressed from the pTATT_BcLF1-Luc reporter (Fig. 7A). This effect of PAA on the amount of luciferase expressed from the pTATT_BcLF1-Luc reporter could be the result of either a toxic effect of the drug on the cells or a decrease in the expression of the endogenous vPIC components due to inhibition of viral DNA replication. In effect, it has been shown, in the CMV model, that expression of some of the early genes whose products are involved in late gene expression are enhanced following viral DNA replication (31). Taken together, these data show that although the vPIC is sufficient for late gene expression in transient assays, viral DNA replication is necessary for late gene expression from the endogenous genome, probably because a change in the structure of the viral chromatin is required before late gene expression is activated.

FIG 7.

Activation of endogenous late gene expression is dependent on viral DNA replication. HEK293_EBV cells were transfected with the pTATT_BcLF1-Luc reporter plasmid together with an EB1 expression vector or the six expression plasmids encoding the vPIC components in the presence or absence of PAA, a specific inhibitor of viral DNA replication. (A) Both EB1 and the vPIC activate luciferase expression from the pTATT_BcLF1-Luc reporter, and this expression is not highly sensitive to PAA treatment. (B) EB1 is unable to activate expression of the endogenous viral late genes in the presence of PAA (50 ng/ml), even when vPIC components are expressed. Following transfection of expression plasmids encoding EB1 or the vPIC components, as indicated in the figure, RNA expression of the endogenous early BMRF1 gene and the late BcLF1 or BDLF1 gene was quantified by RT-qPCR. (C) Control showing complete inhibition of viral DNA replication by PAA (upper panel) and expression of the early BMLF1 and late gp350 proteins analyzed by immunoblotting (bottom panel) in cells used in the experiment whose results are shown in panels A and B of the figure. The experiment was performed 2 times, and the error bars represent standard deviations.

Analogous proteins from HCMV behave like those from EBV.

Since an analogy exists between the EBV components of the vPIC and other beta- and gammaherpesviruses proteins putatively involved in a vPIC, especially HCMV, we tested the effect of a reconstituted HCMV vPIC on the pTATT_BcLF1-Luc reporter. The result of the luciferase assay obtained from HEK293 cells cotransfected with expression vectors for the six HCMV proteins, UL79, UL87, UL91, UL92, UL95, and UL49, was very similar to that of HEK293 cells transfected with the six EBV vPIC protein expression vectors (Fig. 8A). However, none of the HCMV proteins involved in the vPIC could be efficiently replaced with their EBV equivalent (Fig. 8A). The reverse was also true (data not shown).

FIG 8.

Betaherpesviruses also encode a vPIC. (A) HEK293 cells were transfected with the pTATT_BcLF1-Luc reporter plasmid together with the expression vectors encoding either the vPIC EBV components (BcRF1, BFRF2, BGLF3, BVLF1, BDLF4, and BDLF3.5), the vPIC CMV components (UL87, UL49, UL95, UL79, UL92, and UL91), or the vPIC CMV components with one component replaced by the EBV analog, as indicated. The relative luciferase fold activation was calculated relative to the value obtained with the pTATT_BcLF1-Luc reporter transfected alone (−). The experiment was performed 3 times, and the error bars represent standard deviations. Protein expression from transfected cells in panel A was controlled by immunoblotting using a mix of monoclonal and polyclonal anti-Flag antibodies. Neither BDLF3.5 nor UL91 was detected. (B) The CMV vPIC can substitute for the EBV vPIC. HEK293_EBVΔBFRF2 cells were transfected with the different expression plasmids indicated at the bottom of the graph. Total RNA extracted from the transfected cells was quantified by RT-qPCR using primers specific for either the BMRF1 early gene or the BDLF1 and BFRF3 late genes. The experiment was performed 3 times, and the error bars represent standard deviations.

The functional complementation between the EBV and the HCMV vPICs was also demonstrated by transfecting HEK293_EBVΔBFRF2 cells with the complete set of EBV or HCMV vPIC expression vectors and analyzing the expression of the late EBV genes following virus reactivation. As already shown, after virus was reactivated in these cells, the late viral genes BDLF1 and BFRF3 were not expressed, but the early gene BMRF1 was (Fig. 8B, lane 4). Moreover, without reactivation of the viral productive cycle, neither the EBV nor the HCMV vPIC was able to activate expression of the early or late viral genes (Fig. 8B, lanes 2 and 3). As expected, the EBV vPIC efficiently complemented the cells for the expression of the late genes when EBV was reactivated (Fig. 8B, lane 5), and interestingly, the HCMV vPIC gave the same result (Fig. 8B, lane 6).

Taken together, these observations demonstrate that, although each individual EBV vPIC protein cannot be functionally replaced by its HCMV counterpart, the complete HCMV vPIC is able to trans-complement the defect in EBV late gene expression induced by the absence of the EBV vPIC component BFRF2. These results strongly suggest that, as in EBV, the analogous HCMV vPIC proteins form a functional complex that plays a crucial role in the activation of late viral gene expression.

DISCUSSION

Expression of the late genes is an essential step for herpesviruses to complete their productive cycle. Although the initiation of viral late gene expression by viral and/or host factors has remained largely unexplored, previous works have demonstrated that several factors from beta- and gammaherpesviruses, in addition to viral DNA replication, are necessary for the expression of late viral genes. In this study, we have determined the role of the EBV protein BFRF2 during virus infection. We found that when expression of BFRF2 was abolished in cells infected by EBV, the accumulation of early gene products was not affected during the viral productive cycle and viral DNA synthesis proceeded normally. However, viral production was blocked by failure to accumulate late viral transcripts. BFRF2, by supporting the efficient transcription of late viral genes, is absolutely crucial for this accumulation. It appears that the phenotype of the EBV_ΔBFRF2 mutant is reminiscent of that of the EBV_ΔBcRF1 mutant (18), and like BcRF1, BFRF2 has analogs in beta- and gammaherpesviruses. These observations place the BFRF2 protein together with other EBV, CMV, and MHV-68 genes that have been characterized as being essential for late gene expression (18, 26–33).

The BFRF2 protein product consists of two domains with a variable N-terminal part and a C terminus conserved between beta- and gammaherpesviruses. The latter domain contains several conserved cysteine residues, which potentially form a zinc finger protein. Although the detailed activity of BFRF2 needs to be further explored, there are several indications that this protein may be directly involved in binding to DNA.

Studies on beta- and gammaherpesviruses, such as CMV, MHV-68 and EBV, have provided direct insights into the function of a distinct set of conserved genes that are not present in alphaherpesviruses. These genes are called the beta-gamma genes; ORFs 18, 23, 24, 30, 31, 34, and 66 from MHV-68 are analogs of HCMV UL79, UL88, UL87, UL91, UL92, UL95, and UL49 and EBV BVLF1, BcRF1, BTRF1, BDLF3.5, BDLF4, BGLF3, and BFRF2, respectively. All these genes, except for MHV-68 ORF23 (34) and, by analogy, EBV BTRF1 and CMV UL88, appear to be involved in the regulation of late gene expression. Furthermore, we have now demonstrated that the six viral factors interact with one another and that a functional complex comprising these six proteins can be ectopically formed in uninfected cells. This suggests that during EBV infection, a complex containing these viral components forms to promote late gene expression. Moreover, our study provides additional evidence that the regulatory mechanism of viral late gene expression is conserved between EBV and HCMV. The analogous HCMV proteins have the same role in the activation of a late promoter reporter construct. However, it was not possible to substitute any HCMV protein for its EBV analog and vice versa, suggesting that the homology between each protein is not sufficient to allow trans-complementation in the complex. Importantly, however, the complete HCMV complex is able to trans-complement the defect in late gene expression of EBV_ΔBFRF2. Taken together, our results suggest that beta- and gammaherpesviruses encode their own transcription complex responsible for the expression of late viral genes.

The existence of a conserved vPIC between beta- and gammaherpesviruses raises the question of how the alphaherpesviruses regulate their late gene expression without using analogs of the vPIC despite their late promoters being organized in a manner very similar to those of the beta- and gammaherpesviruses. It is interesting to note that for alphaherpesviruses, the viral immediate early proteins ICP4, ICP0, and ICP27 are involved in the expression of late viral genes. In particular, ICP4, which has been shown to be important for the recruitment of the cellular TBP to late gene promoters (36), contains a unique uracil-DNA glycosylase-like domain potentially involved in this process (37). Until now, a direct implication of the CMV or EBV immediate early factors in the regulation of late viral genes has not been described, except for particular genes like that encoding the EBV BLLF1 protein (gp350), whose expression is induced by Rta (6, 38). It is thus tempting to speculate that the EBV BcRF1 protein and its analogs in beta- and gammaherpesviruses play the role of cellular TBP (18). The exact recognition mechanism of minimal promoters in CMV is not known. As previously suggested (17), the CMV complex should not be able to perform a proper molecular recognition of altered TATA box motifs, in coherence with the evolutionary history of CMV promoters, in which no altered consensus TATA box is observed. We are tempted to speculate that the vPIC in CMV is in fact essential for late transcription but needs to be complemented with either cellular or viral factors in order to recognize proper promoters. Additionally, since a differential domain composition of the vPIC elements is observed in beta- and gammaherpesviruses, there is the possibility that these variants contain additional DNA-binding factors responsible for the process of recognition. Based on the conservation of the vPICs between beta- and gammaherpesviruses and the differences with alphaherpesviruses, we can speculate that beta- and gammaherpesviruses have diverged more recently.

A reporter luciferase plasmid driven by a minimal late promoter containing a TATT box was efficiently activated by transient expression of six EBV or HCMV proteins involved in the vPIC in both uninfected and EBV-infected cells not induced into the productive cycle. However, when the productive cycle was induced by transient expression of the viral immediate early transcription factor EB1, activation of the pTATT_BcLF1-Luc reporter was much more important. This difference in the activation level may be the result of the technical difficulty of simultaneously introducing seven expression plasmids in cells by transfection. Thus, the participation of the endogenous vPIC components expressed following induction of the productive cycle by EB1 might explain this difference in activation of the pTATT_BcLF1-Luc reporter. Alternatively, additional accessory viral proteins might be required for the complete activation of the late viral genes. This possibility is reminiscent of what is known for the viral DNA replication complex. Herpesviruses encode a minimum of seven proteins necessary for DNA replication, BALF2, BALF5, BMRF1, BBLF2/BBLF3, BSLF1, BZLF1, and BBLF4 in the case of EBV (7, 39), but additional accessory proteins are also necessary to obtain a more efficient replication complex.

Activation of the BcLF1 and BFRF3 late gene promoters was observed in transiently transfected EBV-infected cells even when the reporter plasmids were devoid of the OriLyt sequence. However, this reconstitution was dependent on EBV DNA replication occurring in the same cells (10), suggesting either that trans-acting factors produced after OriLyt-dependent DNA replication are required for late gene expression or that DNA replication increases the copy number of the viral episome, allowing titration of a putative repressor. In contrast, in a reconstituted stable reporter system, the same late viral promoters showed a specific dependence on OriLyt DNA replication in cis (11), suggesting that during or after DNA replication, the structure of the viral chromatin changes, allowing an increased accessibility to the late promoters for the transcription machinery. If late viral gene transcription is clearly tightly coupled to viral DNA replication, the underlying mechanism is still poorly understood. In accordance with these previous data, we show here that although the vPIC can induce transcriptional activation of the viral late genes in a transient assay, it is completely inefficient at inducing expression of the endogenous late genes from the viral genome in the absence of viral DNA replication. These results may suggest that the vPIC controls late viral gene transcription in association with the viral DNA replication complex, although there is currently no direct evidence that any of the beta-gamma gene products interact with proteins from the viral DNA replication complex. However, Sugimoto et al. (40) have reported that the EBV BcRF1 protein localizes inside the replication foci (BMRF1 cores) in late stages of lytic replication, suggesting that late gene expression is initiated from the newly replicated DNA in the replication foci.

How the vPIC functions and which additional protein components are present in this complex remain important questions. We can speculate, as previously suggested (32), that the vPIC may activate late gene transcription by remodeling the chromatin structure of the viral genome. Genomes of Herpesviridae associate with histones during infection and require epigenetic regulation/modification for gene expression (41–43). However, in contrast to the original episomic viral genome, the newly synthesized DNA is probably not covered by nucleosomes (44). This point may explain why, in transient transfection, it is possible to activate late viral gene transcription, whereas this is not possible with the endogenous viral genome. Alternatively, and more likely, the vPIC may play a more direct role in linking viral DNA replication and transcription. We have previously shown that BcRF1, the homolog of CMV UL87 and MHV-68 ORF24, acts as a TATA-binding protein, preferentially interacting with TATT sequences, to regulate late gene expression during infection (17, 18). Furthermore, it has been shown that MHV-68 ORF30 and ORF34, homologs of CMV UL91 and UL95 and EBV BDLF3.5 and BGLF3, respectively, regulate the binding of RNAP II to the late gene promoters (27). All this evidence points toward a direct contribution by the beta-gamma late gene regulators in the assembly of a late gene RNAP II transcription complex, a multicomponent enzyme that requires a host accessory scaffold and regulatory proteins for its activity. The vPIC may play an essential role in the recruitment of these components to or their assembly on viral late gene promoters as cellular TBP does for the recruitment of TFIID and the cellular preinitiation complex. In accordance with this, we have shown that BcRF1 is associated with the hyperphosphorylated form of RNAP II. Further studies on the formation of this viral complex as well as its regulation and recruitment to the TATT element will be an important basis for dissection of the mechanism of beta- and gammaherpesvirus late gene regulation. How the vPIC interacts with viral and/or host factors and its relationship to core components of the RNAP II transcriptional machinery as well as viral DNA synthesis will be the subject of future investigations.

This study advances our understanding of the mechanisms of EBV late gene expression, an important question that has long remained unresolved. Furthermore, this very specific vPIC, present in beta- and gammaherpesviruses and required for productive infection, is potentially a very attractive target for antiviral strategies to combat HCMV, KSHV, and EBV infection as well as their reactivation and associated diseases.