An Epstein-Barr Virus-Encoded Protein Complex Requires an Origin of Lytic Replication In Cis to Mediate Late Gene Transcription

Epstein-Barr virus lytic replication is accomplished by an intricate cascade of gene expression that integrates viral DNA replication and structural protein synthesis. Most genes encoding structural proteins exhibit “true” late kinetics–their expression is strictly dependent on lytic DNA replication. Recently, the EBV BcRF1 gene was reported to encode a TATA box binding protein homolog, which preferentially recognizes the TATT sequence found in true late gene promoters. BcRF1 is one of seven EBV genes with homologs found in other β- and γ-, but not in α-herpesviruses. Using EBV BACmids, we systematically disrupted each of these “βγ” genes. We found that six of them, including BcRF1, exhibited an identical phenotype: intact viral DNA replication with loss of late gene expression. The proteins encoded by these six genes have been found by other investigators to form a viral protein complex that is essential for activation of TATT-containing reporters in EBV-negative 293 cells. Unexpectedly, in EBV infected 293 cells, we found that TATT reporter activation was weak and non-specific unless an EBV origin of lytic replication (OriLyt) was present in cis. Using two different replication-defective EBV genomes, we demonstrated that OriLyt-mediated DNA replication is required in cis for TATT reporter activation and for late gene expression from the EBV genome. We further demonstrate by fluorescence in situ hybridization that the late BcLF1 mRNA localizes to EBV DNA replication factories. These findings support a model in which EBV true late genes are only transcribed from newly replicated viral genomes.


Introduction
Epstein-Barr virus (EBV) is a γ-herpesvirus that infects more than 95% of the human adult population. Primary infection is usually asymptomatic if acquired early in life, but often results in infectious mononucleosis in adolescence [1,2]. EBV is associated with several B-cell and epithelial cancers, including Burkitt lymphoma, Hodgkin lymphoma, post-transplant lymphoproliferative disorder, nasopharyngeal and gastric carcinoma [3][4][5][6]. Like all herpesviruses, EBV exists in two states in infected cells: latent infection and lytic replication. In latent infection, only a limited subset of viral genes is expressed, and infectious virions are not produced. In contrast, during lytic replication (or productive infection), nearly all viral genes are transcribed, viral DNA is replicated, and infectious virions are produced, enabling transmission to other cells and hosts. Although EBV-positive tumors are characterized by latent infection, there is increasingly compelling evidence that both latent infection and lytic gene expression are essential for emergence of EBV-associated malignancies [7][8][9][10].
As with all herpesviruses, EBV lytic replication proceeds through an ordered cascade of gene expression. First, immediate-early genes BRLF1 and BZLF1, encoding the transcription factors R (also called Rta) and Z (also called Zta, ZEBRA or EB1), respectively, are expressed [11][12][13][14][15][16]. R and Z activate the promoters of early genes [12,[17][18][19]. Many early genes encode proteins directly or indirectly involved in viral DNA replication, which is initiated at two origins of lytic replication (OriLyt) used during the productive cycle of EBV [20]. Following viral DNA synthesis, late genes are expressed. EBV late genes mainly encode structural proteins of the EBV virion that elicit strong immune responses (reviewed in reference [11]). EBV late genes can be further subdivided into two groups based on the degree of dependence on viral DNA replication. True late genes absolutely require viral DNA replication for their expression, while leaky late (or delayed early) genes, although usually expressed after DNA replication, can still be expressed from viruses defective for DNA replication. True late genes are often distinguished by their inability to be expressed in presence of viral DNA polymerase inhibitors such as acyclovir [21]. While regulation of immediate-early and early gene expression in herpesviruses is well characterized, relatively little is known about regulation of late genes.
In contrast to all other viral gene promoters, the promoters of EBV true late genes (simply referred to as late genes hereafter) frequently contain a non-canonical TATA box with a thymidine at the fourth position (TATT) [22]. This non-canonical TATT sequence has been recently identified as a critical element for activation of late genes in the β-herpesvirus human cytomegalovirus (HCMV), as well as other γ-herpesviruses, including Kaposi's sarcoma-associated herpesvirus (KSHV) and murine herpesvirus-68 (MHV-68) [23][24][25]. An important advance in understanding of EBV late gene regulation came with the prediction that EBV BcRF1 encodes a TATA box binding protein (TBP)-homolog that preferentially recognizes the TATT boxcontaining promoters [26]. This prediction was subsequently confirmed experimentally; furthermore, deletion of BcRF1 from the EBV genome was found to inhibit late gene expression without impairing lytic DNA replication [27]. BcRF1 is one of 7 genes (BcRF1, BDLF3.5, BDLF4, BFRF2, BGLF3, BTRF1, and BVLF1) with homologs found in βand γbut not in αherpesviruses (referred to as βγ genes hereafter). Interestingly, in both MHV-68 and HCMV, 5 out of 7 βγ genes, including the BcRF1 homologs, have been reported to produce the identical phenotype when deleted: 1) intact viral DNA synthesis; and 2) impaired late gene expression [28][29][30][31][32][33][34][35]. In EBV, two additional βγ gene knockout viruses deleted for BDLF4 and BFRF2 have been characterized and found to exhibit this specific defect in late gene expression [36,37]. Additionally, in reporter assays, 6 EBV βγ genes (all but BTRF1) were required together to activate a TATT-driven luciferase construct in EBV-negative 293 cells [37]. The requirement for BDLF3.5, BGLF3, BTRF1, and BVLF1 for late gene expression in the context of the viral genome is unknown.
The existence of a strict link between late gene transcription and the onset of viral DNA replication is observed among many DNA viruses. In herpesviruses, a critical unresolved question in understanding this link is whether an origin of lytic replication is required in cis with the late gene or if DNA replication in trans is sufficient to permit late gene expression. For example, in SV40, DNA replication in trans is sufficient to titrate transcriptional repressors that block late gene expression from the SV40 genome at early stages of replication [38]. In contrast, a requirement for an OriLyt in cis implies the parental genome is not competent for late gene transcription and that newly replicated DNA serves as the template for production of late gene mRNAs.
Attempts to resolve whether OriLyt is required in cis or in trans have met with conflicting results, in part because studies using plasmids do not always recapitulate effects in the context of whole viral genomes [39]. In the case of MHV-68, it was shown that a reporter plasmid containing a late promoter could not be activated unless OriLyt was present on the plasmid [25]. Studies in KSHV suggest that activation of KSHV K8.1 late promoter can be enhanced by the left-end viral origin of replication, OriLyt-L, largely in trans [24]. The requirement of OriLytmediated DNA replication in cis or in trans for late gene expression is also controversial in the EBV field. Using a transient late promoter-reporter system, Serio et al. showed that although activity of such reporters required EBV lytic replication occurring in the same cell, OriLytmediated DNA replication was not required in cis [22]. In contrast, Amon et al. demonstrated that robust late gene expression was detected in EBV-positive cell lines stably expressing latepromoter reporter plasmids only when EBV OriLyt was present on the reporters [40]. Finally, studies by Aubry et al. describe a viral pre-initiation complex (vPIC) encoded by 6 βγ genes (BcRF1, BDLF3.5, BDLF4, BFRF2, BGLF3, and BVLF1) that activates a TATT-containing promoter reporter plasmid lacking an OriLyt in EBV negative cells [37]. Thus, the requirement for OriLyt-mediated DNA replication for EBV late gene expression appears to depend upon the model system employed and has not been examined in the context of the intact EBV genome.
Here, we used EBV BACmids to analyze the role of each βγ gene in late gene expression. Using cell lines stably infected with EBV BACmids individually mutated for each βγ gene, we demonstrate that 6 out of 7 (BcRF1, BDLF3.5, BDLF4, BFRF2, BGLF3, and BVLF1) βγ genes are required for late gene expression from the virus, but dispensable for DNA replication. In reporter assays, βγ gene mediated activation of late-gene promoters required thymidine at the forth position of the non-canonical TATT sequence and, importantly, strictly required OriLytmediated DNA replication. Furthermore, using replication-defective EBV BACmids (lacking the single-stranded DNA binding protein BALF2 or OriLyt), we found that OriLyt-mediated DNA replication in cis, and not in trans, is a pre-requisite for EBV late gene expression. Finally, using fluorescence in situ hybridization and a visible EBV derivative, we demonstrate that the BcLF1 late mRNA localizes to the EBV replication factories. Our results are consistent with a model in which the viral pre-initiation complex encoded by 6 EBV βγ genes mediate late gene transcription from newly replicated viral DNA.

Results
Six of the seven EBV βγ genes are essential for late gene expression, but dispensable for early gene expression To investigate the requirement of each of the 7 βγ genes for early and late gene expression, we derived 293 cell lines infected with EBV BACmids with one βγ gene mutated. We constructed the ΔBDLF3.5, ΔBGLF3, and ΔBTRF1 BACmids using the En Passant method [41,42] and obtained WT, ΔBcRF1 (MI-27), ΔBDLF4 (MI-84), ΔBFRF2 (MI-248) and ΔBVLF1 (MI-383) BACmids from a library of mutant EBV BACmids previously described [43]. Mutations were confirmed by sequencing of high fidelity PCR products from the appropriate region of each BACmid. Integrity of each BACmid was preliminarily assessed by restriction digestion using at least two enzymes (BamHI and EcoRI) and subsequently confirmed by trans-complementation (discussed below). 293 cells were infected with each EBV mutant BACmid, and several singlecell clones were selected for further validation. Each EBV Δβγ 293 line was induced for lytic replication by transfection with R and Z expression plasmids and trans-complemented with a plasmid expressing the missing βγ gene, where indicated. We found that 6 out of 7 βγ genes (BcRF1, BDLF3.5, BDLF4, BFRF2, BGLF3, and BVLF1) were required for expression of the minor capsid protein VCAp18 (product of the EBV BFRF3 late gene) at 72 hours post induction with R and Z ( Fig 1A). In contrast, EBV BTRF1 was dispensable for VCAp18 expression (Fig 1B). None of the 7 βγ gene knockout genomes exhibited a defect in expressing the protein product of the BMRF1 early gene (Fig 1A and 1B). These results confirm and extend previous studies that demonstrated BcRF1, BFRF2, and BDLF4 share a common phenotype: defective late gene expression with intact early gene expression [27,36,37]. We have now shown that this same phenotype is shared by 3 other βγ genes: BDLF3.5, BGLF3, and BVLF1, but not by BTRF1, the last of the 7 genes conserved in all βand γ-herpesviruses.
βγ genes are dispensable for viral DNA replication Because late gene expression is dependent on lytic viral DNA replication, it was important to determine whether any of the βγ genes played a role in viral DNA replication. For these experiments, we induced each of the six EBV Δβγ 293 lines, which exhibited a defect in VCAp18 expression, for lytic replication by transfection of R and Z with or without βγ gene trans-complementation. We then measured EBV DNA (relative to cellular GAPDH) by qPCR. As demonstrated in Fig 2, a 100-fold or greater increase in EBV DNA was observed in the EBV ΔBcRF1, ΔBDLF3.5, ΔBDLF4, ΔBFRF2, ΔBGLF3, and ΔBVLF1 293 cells in response to R and Z expression. No further increase in EBV DNA was observed upon βγ trans-complementation, consistent with BcRF1, BDLF3.5, BDLF4, BFRF2, BGLF3, and BVLF1 playing no role in supporting EBV DNA replication. Our results are consistent with prior studies demonstrating that BALF2, BALF5, BBLF2/3, BBLF4, BMRF1, BSLF1, BKRF3, Z, R, and SM are sufficient to reconstitute EBV lytic DNA replication from a lytic origin in vitro [44,45]. Our results further demonstrate that none of the 6 βγ genes tested play a role in EBV lytic DNA replication that could account for their role in late gene expression.

An EBV OriLyt in cis is required for activation of a TATT-containing reporter
In order to determine whether lack of TATT promoter activation correlated with the late gene expression defect observed in our βγ gene knockout genomes, we constructed a reporter, pGL2-TATT similar to the one described by Gruffat et al. [27], in which the promoter of the late gene BcLF1 (containing the non-canonical TATT) drives expression of luciferase. A corresponding control reporter, pGL2-TATA, was constructed by mutation of the fourth T to an A, resulting in a conventional TATA box. We performed reporter assays, with these reporters, using 293 cells infected with an EBV genome defective for R expression (EBV R-stop [46]) that exhibits no spontaneous lytic replication in the absence of transfected R. In this model, we observed an approximately 40-fold activation of the pGL2-TATT reporter in response to Rinduced EBV replication ( Fig 3A) which decreased to about 15-fold upon addition of acyclovir. Unexpectedly, the control pGL2-TATA was also activated about 35-fold, suggesting much of the observed activation was TATT independent. We hypothesized that EBV OriLyt may be required for specific activation of TATT promoter. To test this possibility, we cloned the EBV OriLyt sequence downstream of the luciferase gene to generate pGL2-TATT-OriLyt. To evaluate the OriLyt requirement for βγ gene mediated activation of TATT-containing promoters, each EBV Δβγ 293 line was transfected with either the pGL2-TATT or pGL2-TATT-OriLyt reporter plasmids, and induced with R and Z with or without trans-complementation of the disrupted βγ gene. As shown in Fig 3B, the pGL2-TATT-OriLyt reporter was robustly induced (30-500 fold) in each cell line. Only weak activation of the pGL2-TATT plasmid was observed, typically at levels 1-10% of that observed for pGL2-TATT-OriLyt. Activation of the pGL2-TATT-OriLyt construct was strongly dependent on trans-complementation with the missing βγ gene in each cell line. Furthermore, because OriLyt-mediated DNA replication is unaffected by βγ gene trans-complementation (Fig 2), it is unlikely that the OriLyt dependence of the βγ gene activity is due solely to an increase in reporter copy number. Thus, 6 out of 7 EBV βγ genes (BcRF1, BDLF3.5, BDLF4, BFRF2, BGLF3, and BVLF1) contribute to activation of a TATT reporter plasmid and the presence of the EBV OriLyt is required for high-level activation.
We also examined the requirement for OriLyt from a different late promoter using the pTATT-GFP-VCAp18-OriLyt and pTATT-GFP-VCAp18 reporter plasmids which express a GFP-VCAp18 fusion protein from the native VCAp18 (BFRF3) promoter with or without the EBV OriLyt sequence cloned 3' to the expression cassette, respectively. Both plasmids have an EBV OriP latent origin, permitting their maintenance as an episome in EBV-infected cells. Initially these reporters were transfected into EBV WT 293 cells with or without R and Z expression plasmids to induce lytic replication. Cells were subjected to microscopy at 24, 48, and 60 hours after transfection to test for expression of GFP-VCAp18. As shown in Fig    We further evaluated the requirements of βγ gene mediated late gene transcription by constructing an additional reporter, pTATA-GFP-VCAp18-OriLyt, in which the BFRF3 promoter TATT sequence was mutated to the canonical TATA. As shown in  BcRF1-mediated transcription was not supported by reporters lacking OriLyt (middle panels) or bearing the TATT to TATA mutation (bottom panels). Results using 293 cells infected with EBV ΔBDLF3.5, EBV ΔBDLF4, EBV ΔBFRF2, EBV ΔBGLF3, and EBV ΔBVLF1 were similar: GFP-VCAp18 expression was only observed from the pTATT-GFP-VCAp18-OriLyt reporter and required induction of viral replication plus trans-complementation of the absent βγ gene. Collectively these results suggest that late gene promoter activation requires each of 6 βγ genes, the non-canonical TATT-box, and the EBV origin of lytic replication.

Construction of an EBV mutant BACmid defective for lytic viral DNA replication
Because the EBV OriLyt functions as both a replication origin and a transcriptional enhancer, it was important to determine whether or not the requirement for OriLyt on the reporter reflected a need for the reporter plasmid DNA to be replicated in order to activate the TATT element. To assess this requirement for newly replicated DNA, we constructed a DNA replication-defective EBV BACmid by inserting a stop codon in place of the second amino acid codon of the BALF2 gene which encodes the single stranded DNA binding protein, an essential component of the viral DNA replication machinery. This BACmid had previously been modified by insertion of a sequence encoding an N-terminal HA epitope tag upstream of the BcRF1 gene and therefore was designated as EBV ΔBALF2/HA-BcRF1. 293 cells were stably infected with EBV ΔBALF2/ HA-BcRF1, then induced with R and Z and trans-complemented with a plasmid expressing BALF2, where indicated. Cells were harvested 48 hours post induction and qPCR was performed with primers specific to EBV OriLyt and the cellular GAPDH. As shown in Fig 5A, when the BALF2 gene is mutated, viral DNA replication does not occur. This defect is rescued when BALF2 is expressed in trans. Immunoblotting revealed that EBV ΔBALF2/HA-BcRF1 239 cells express the BMRF1 early gene product in the absence of BALF2, but that expression of the VCAp18 late gene product is strictly dependent on BALF2 trans-complementation ( Fig 5B). Furthermore, loss of BALF2 did not have an effect on relative βγ mRNA levels (S1 Fig). The replication-defective EBV ΔBALF2/HA-BcRF1 293 cells, therefore, provide an appropriate system for assessing the requirement of viral DNA synthesis for late gene expression, while bypassing the potential cytotoxic and/or off-target effects of the herpesvirus DNA polymerase inhibitor, acyclovir.

Viral DNA replication is necessary for TATT promoter activation
To understand the role of viral DNA replication for late gene expression, we measured TATT promoter activity in 293 cells containing the replication-defective EBV BACmid described in Fig 5. EBV ΔBALF2/HA-BcRF1 293 cells were transfected with the pGL2-TATT and pGL2-TATT-OriLyt (containing the promoter of the late gene BcLF1) reporter plasmids and induced with R and Z in presence or absence of BALF2 trans-complementation. Cells were harvested 48 hours post lytic induction and reporter assays were performed. As shown in Fig 6A, the TATT promoter can only be robustly activated when the EBV OriLyt is present on the plasmid and viral DNA replication is rescued via BALF2 trans-complementation. This is consistent with the requirement of OriLyt-mediated viral DNA synthesis for TATT promoter activation.
To further confirm the requirement of viral DNA synthesis for TATT activation and late gene expression, we examined expression from the pTATT-GFP-VCAp18-OriLyt reporter in the DNA replication-defective EBV ΔBALF2/HA-BcRF1 infected 293 cells. Cells were induced with R and Z, trans-complemented with a plasmid expressing BALF2 to rescue the replication defect where indicated, and assessed for GFP positivity at 24, 48, and 60 hours post induction by microscopy. GFP-positive cells were only observed at 60 hours when viral DNA replication was rescued by BALF2 trans-complementation (Fig 6B). No GFP-positive cells were observed at  24 or 48 hours post induction, consistent with late kinetics of GFP-VCAp18 expression. These observations are in accord with the hypotheses that OriLyt-mediated DNA replication from the pTATT-GFP-VCAp18-OriLyt reporter is required for GFP-VCAp18 expression to occur.

EBV TATT-driven late gene expression requires OriLyt in cis
Although we found a dependence on OriLyt-mediated DNA replication in two distinct reporter systems, it was important to determine if this was required for late gene expression in the context of the EBV genome. Therefore, we first deleted the remaining OriLyt from the EBV WT BACmid (one OriLyt is absent because the BACmid is derived from the B95.8 EBV strain) and established stable 293 cell lines. pTATT-GFP-VCAp18-OriLyt or pTATT-GFP-VCAp18 were transfected into 293 cells stably infected with EBV WT or EBV ΔOriLyt and the lytic cycle was induced by transfection of R and Z expression plasmids (Fig 7A). We then performed immunoblot assays to measure VCAp18 from the endogenous EBV genome or GFP-VCAp18 from the exogenous plasmid in the presence or absence of an OriLyt in cis. Cells were harvested 60 hours post induction and immunoblot assays were performed using antibodies against the early protein BMRF1 and the late protein VCAp18. VCAp18 signal from the endogenous EBV genome was detected in the context of WT EBV, but not with EBV ΔOriLyt genome even when OriLyt was present in trans on the pTATT-GFP-VCAp18-OriLyt plasmid (Fig 7B). Likewise, GFP-VCAp18 was expressed from pTATT-GFP-VCAp18-OriLyt but not from pTATT-GFP-VCAp18, even when an OriLyt was present on the WT EBV genome in trans. Thus, the late protein VCAp18 is only expressed when OriLyt is present in cis, but the early protein BMRF1 is expressed equally well in the presence or absence of OriLyt. Thus, we conclude that OriLyt-mediated DNA replication is specifically required in cis for EBV late gene expression from TATT promoters.
The BcLF1 late mRNAs co-localize exclusively with EBV DNA replication factories It has been previously shown that EBV lytic DNA replication occurs in nuclear compartments or "factories" that are devoid of histones and cellular DNA [47]. The requirement for an OriLyt in cis suggests a functional link between these factories and late gene transcription. Using a previously described "visible" EBV derivative that contains 250 lacO binding sites and expresses a LacI-tdTomato fluorescent fusion protein, we identified DNA replication factories and sought to determine if they co-localized with late mRNAs. When cells were imaged at 36 hours, early DNA replication factories were observed in some cells (Fig 8A). In these cells, fluorescent in situ hybridization (FISH) for the early BALF2 mRNA revealed both a nuclear and cytoplasmic distribution. At 48 hours post-induction, EBV DNA replication factories were evident as large foci of red fluorescence (Fig 8B, Visible EBV panels). Detection of the BcLF1 mRNA at this time ( Fig 8B, late mRNA panels) by FISH showed punctate staining that co-localized with sites of DNA replication (Fig 8B, merge). These results demonstrate that late mRNAs synthesis corresponds temporally and geographically with sites of productive EBV DNA synthesis.

Discussion
A growing body of evidence supports the hypothesis that βand γ-herpesviruses control late gene expression via a virus-encoded pre-initiation complex [26][27][28][29][30][31][32][33][34][35][36][37]. Here, we analyzed the phenotype of EBV mutants with disruptions in each of the seven genes with homologs in βand γ-herpesviruses (βγ genes). We found that 6 of the 7 βγ genes are essential for expression of the EBV VCAp18 late gene product and dispensable for lytic DNA replication (Figs 1 and  2). These results are consistent with recently published reports assessing the roles of the BcRF1, BFRF2, and BDLF4 genes [27,36,37]. We extended these findings, demonstrating here that this phenotype is also true for three additional EBV βγ genes (BDLF3.5, BGLF3, and BVLF1), but not for the last EBV βγ gene, BTRF1. These same six proteins, when exogenously expressed in EBV-negative 293 cells, have been reported to form a viral pre-initiation complex that mediates transcription from the non-canonical TATT box found in late gene promoters [37]. Unexpectedly, using 293 cells infected with EBV genomes deleted for specific βγ genes, we found that βγ gene-mediated activation of a TATT reporter was weak and non-specific (relative to TATA) unless an EBV OriLyt was present on the reporter (Figs 3 and 4). Although the EBV OriLyt acts as both an enhancer and an origin of DNA replication, we demonstrated that the TATT-OriLyt reporter could not be activated in 293 cells infected with a replication defective EBV genome. Thus, lytic DNA replication is required for optimal EBV-mediated activation of the TATT reporter. We further demonstrated, using a combination of OriLyt deleted EBV genomes and plasmids, that an OriLyt in cis is essential for TATT-driven late gene expression (Fig 7). Finally, we demonstrated that the late BcLF1 mRNA exclusively localizes to the EBV replication factories; In contrast, the early BALF2 mRNA is more broadly distributed (Fig 8).
Based on these results we propose a model where βγ gene products mediate recruitment of RNA polymerase II to transcribe EBV late genes from newly replicated viral DNA (illustrated in Fig 9).
In an effort to most closely model EBV late gene expression, our experiments presented here were performed in cells infected with EBV WT or EBV mutant genomes. This approach offers multiple advantages over reporter assays in EBV-negative cells. First, EBV lytic gene products required for late gene expression are expressed under the control of the native promoters assuring proper temporal kinetics and levels of expression. Additionally, because the extent to which βγ gene products are modulated by other EBV lytic proteins (e.g., the BGLF4 kinase) remains to be determined, it is preferable to study their role in the presence of the EBV genome. Finally, late gene expression via βγ gene products has been compared to the T4 bacteriophage late gene regulation [34]. It is interesting to note that for T4, DNA nicks bypass the dependence of late gene expression on DNA replication [48]. If this holds true for EBV, one would expect the apparent requirement for OriLyt in reporter assays to be strongly influenced by the transfection conditions and the integrity of the reporter plasmid. For these reasons, we have endeavored to make or confirm all the observations reported here in EBV infected cells with plasmids that can be maintained extrachromosomally.
Why would an OriLyt in cis be required for late gene mediated transcriptional activation? The full EBV OriLyt is both a strong enhancer and a DNA replication origin. Indeed, transcriptional activation and the subsequent formation of RNA-DNA hybrids at OriLyt appear to be essential for initiation of lytic DNA replication in γ-herpesviruses (as well as cytomegalovirus) which lack dedicated origin binding proteins [49,50]. Because a minimal OriLyt exhibiting 1% of the DNA replication activity of full OriLyt supports late gene expression [40], it has been suggested that other mechanisms such as OriLyt dependent PML body association may be important [51]. Our results do not exclude these additional mechanisms; however, our ΔBALF2 experiment (Fig 6) and the ability of viral DNA polymerase inhibitors to block late gene expression argue that the requirement for OriLyt includes its DNA replication function. Further, we believe the linkage of βγ mediated late gene transcription to the requirement for an OriLyt in cis, strongly implies that the parental EBV genome is not competent for late gene transcription until it has been replicated.
In EBV, the viral replication machinery consisting of the BALF5 DNA polymerase, BMRF1 DNA polymerase processivity factor, BALF2 single-stranded DNA binding protein, and the BBLF4-BSLF1-BBLF2/3 helicase-primase complex replicate the EBV genome during productive infection in discrete nuclear sites called factories. Replication factories form as a consequence of a dramatic rearrangement in nuclear architecture, in which the cell DNA moves to the nuclear periphery and the nucleus becomes dominated by the expanding replication factory. These factories are characterized by the presence of newly synthesized viral DNA and the EBV DNA polymerase and its processivity factor, BALF5 and BMRF1 respectively, but are devoid of cellular DNA, histones, and PCNA [47]. Other investigators have used BMRF1 immunofluorescence to define EBV replication factories as BMRF1 cores [52,53] that lend additional support to our model that late gene mRNAs are transcribed from newly replicated viral DNA. Using pulse-chase experiments, Sugimoto et al. demonstrated that newly synthesized viral genomes organized around the BMRF1 cores were transferred inward, leaving the parental template outside the BMRF1 cores [53]. Using a "visible" EBV derivative, we demonstrated that the BcLF1 late mRNA co-localizes exclusively with DNA replication factories, in contrast the BALF2 early mRNA which was predominantly cytoplasmic at the time points studied and often identified in cells that had not formed DNA replication factories. Interestingly, BcRF1, the EBV TATT-binding protein, is primarily localized inside the BMRF1 cores where newly replicated DNA is localized [54], and therefore is well positioned to interact with newly synthesized viral DNA. Although not definitively demonstrated, these findings collectively suggest that EBV, and by analogy other βand γherpesviruses, employ a set of evolutionary conserved proteins to mediate transcription of late genes from the newly replicated DNA template.
Although the dependence of late gene expression on viral DNA replication is observed in all herpesviruses, our work adds to an increasing body of evidence that βand γherpesviruses regulate their late genes by a distinct mechanism. α-herpesvirus late genes are similar to βand γ-herpesvirus late genes in that their expression is controlled by a short proximal TATA box [55], but differ in that this sequence neither appears to be similar to the unconventional TATT box found in βand γ-herpesviruses, nor is it recognized by unique virally encoded trans-acting factors. In herpes simplex virus (HSV), a limited number of genes such as ICP4, ICP8 and ICP27 have been implicated [56][57][58][59][60][61][62][63][64], but their disruption impairs other stages of HSV replication as well as late gene expression. The preponderance of evidence suggests that βand γ-herpesviruses encode six conserved gene products that mediate late gene expression by recruiting RNA polymerase II to TATT elements in late gene promoters. Our demonstration that OriLyt is required in cis is particularly appealing as it further explains the dependence of late gene expression on lytic DNA replication. That α-herpesviruses recapitulate this phenotype with a mechanistically distinct regulatory apparatus suggests that strong convergent evolutionary pressure exists to regulate late gene expression in a DNA replication-dependent fashion.
What are the potential evolutionary advantages of a viral pre-initiation complex linking late gene expression to DNA replication? One possibility is that such a system provides direct means for expressing structural protein mRNAs in direct proportion to the quantity of replicated viral DNA available for packaging. Such a system may also contribute to viral immune evasion by ensuring that regardless of the transcription factor milieu of the infected cell, a large number of potential antigens cannot be expressed unless the virus has committed to replicating its DNA. In a similar vein, by obviating the need for cell transcription factors to promote late gene expression, there is less potential for interference or cross-talk among viral lytic promoters. Given the propensity of α-herpesviruses to establish latency in neurons, it is tempting to speculate that some or all of these advantages facilitated the success of βand γ-herpesviruses in establishing latency in cells with potential for dynamic changes in transcriptional regulation such as lymphocytes and hematopoietic stem cells.
If indeed βand γ-herpesviruses direct RNA polymerase II to transcribe newly replicated DNA, it raises many interesting questions since mRNA transcription is normally suppressed during DNA replication. It is likely that the βγ pre-initiation complex must overcome a number of cellular checkpoints designed to prevent such transcription. Because nascent viral DNA is thought to be devoid of histones [47], proteins that bind specific histone modifications would need to be recruited by other means. Some progress has been made in mapping the protein-protein interactions that allow the βγ gene products to assemble into a pre-initiation complex [65,66]. At present, we only understand the specific role of one of these βγ genes, the TBP homolog-encoded BcRF1. The BDLF4 gene product and its paralogs are notable for a probable zinc coordination motif (Cys-X 2 -Cys-X 3 -His-X-Cys-X 5,6 -Cys-X 10 -Cys) in their N-termini that could mediate DNA or protein interactions. It will be important to determine the functions of the remaining βγ gene products and understand how they act to permit transcription from this atypical DNA template.

Cell lines and culture
All cell lines were maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. EBV-negative 293 cell lines were obtained from Bill Sugden, University of Wisconsin-Madison. 293 cells infected with the EBV R-stop mutant (a gift from Shannon Kenney, University of Wisconsin-Madison) have been previously described [46].

Reporter assays
Cells were transfected in 12-well plates with 0.25 μg of reporter plasmid, 0.25 μg of trans-complementing plasmid where indicated, 0.02-0.125 μg of R and Z expression plasmid, and 0.025 μg of a plasmid expressing renilla luciferase (as a control) using Lipofectamine 2000 (Invitrogen) according to manufacturer's protocol. After 48 hours, cells were lysed in passive lysis buffer (Promega) for 15 minutes at room temperature on a rocking platform and clarified by centrifugation. Firefly and renilla luciferase values were measured using the dual-luciferase reporter assay kit (Promega) on a BD Monolight 3010 luminometer (BD Biosciences).

Quantification of viral DNA copy number by real-time (RT) PCR
EBV-positive 293 cells were induced in 12-well plates using 125 ng each of R and Z expression plasmids along with 250 ng of the trans-complementing plasmid. 48 hours post lytic induction, cells were washed with phosphate-buffered saline (PBS), and genomic DNA was extracted using GeneJET genomic DNA purification kit (Thermo Scientific) according to manufacturer's protocol.
Single-molecule RNA fluorescence in situ hybridization RNA fluorescence in situ hybridization (FISH) was performed following manufacturer's protocol (LGC Bioresearch Technologies). The probes employed are single-stranded DNA oligos (20 nucleotides), each labeled with the fluorophore Quasar 670 and were designed using online Stellaris probe designer provided by manufacturer (LGC Bioresearch Technologies). i293 Visible EBV cells [47] were plated on coverslips and grown overnight at 37°C.

RNA isolation, reverse transcription and quantification by real-time (RT) PCR
EBV-positive 293 cells were induced in 12-well plates using 125 ng each of R and Z expression plasmids along with 250 ng of the trans-complementing plasmid. 48 hours post lytic induction, cells were washed with phosphate-buffered saline (PBS), and RNA was extracted using Gene-JET RNA purification kit (Thermo Scientific) according to manufacturer's protocol with the following modification: after lysis and before loading on column, lysates was passed through a QIAshredder cell and tissue homogenizer (Qiagen). The eluted RNA was then treated with DNase (1 unit/μg DNA), DNase was deactivated by incubation at 65°C and the treated RNA (~1 μg) was reverse transcribed using the ImProm-II Reverse Transcription System (Promega). Purified cDNA was subjected to qRT-PCR with a 7900HT Fast Real-Time PCR system (Applied Biosciences) using SYBR Green Real-Time PCR Master Mix (Biorad). The primers used to detect various EBV transcripts and β-Actin are listed in S1 Table. Supporting Information (TIF) S1 Table. Primers used for detection of cDNAs corresponding to 6 essential βγ transcripts and the β-Actin reference. (PDF)