The Plasmid Replicon of Epstein-Barr Virus: Mechanistic Insights Into Efficient, Licensed, Extrachromosomal Replication in Human Cells

Abstract

The genome of Epstein-Barr Virus (EBV) and plasmid derivatives of it are among the most efficient extrachromosomal replicons in mammalian cells. The latent origin of plasmid replication (oriP), when supplied with the viral Epstein-Barr Nuclear Antigen 1 (EBNA1) in trans, provides efficient duplication, partitioning and maintenance of plasmids bearing it. In this review, we detail what is known about the viral cis and trans elements required for plasmid replication. In addition, we describe how the cellular factors that EBV usurps are used to complement the functions of the viral constituents. Finally, we propose a model for the sequential assembly of an EBNA1-dependent origin of DNA synthesis into a pre-Replicative Complex (pre-RC), which functions by making use only of cellular enzymatic activities to carry out the replication of the viral plasmid.

1. General Background

Epstein-Barr Virus (EBV), a member of the herpesvirus family, is a surprisingly successful parasite, as are other human members of this family of viruses. It establishes a persistent, lifelong infection in greater than 90% of the world’s population (Fields, Knipe et al. 2006). Primary infection by EBV causes infectious mononucleosis in some, usually adolescent people (Evans, Niederman et al. 1968). The latent cycle of the virus that follows infection is usually asymptomatic in the host, however in a small fraction of infected people, EBV contributes to several cancers including Burkitt’s lymphoma, some T-cell lymphomas, Hodgkin’s disease, post-transplant lymphoproliferative disease, nasopharyngeal carcinoma, and gastric carcinoma. EBV was the first human virus to be classified as a tumor virus (reviewed in Chapter 75 of (Fields, Knipe et al. 2006)) and has been studied because of its association with these diseases. EBV, in addition, has also served as a model to study DNA replication in human cells because of it’s ability both to maintain its genome as an extrachromosomal replicon in infected B-cells, and to appropriate the cellular DNA replication machinery needed for its licensed replication.

2. Cis Elements of EBV that Contribute to DNA Replication

The genome of EBV is approximately 165kbp (Bloss and Sugden 1994) and resides as a nucleosome-coated nuclear plasmid in infected cells (Wensing and Farrell 2000), (Shaw, Levinger et al. 1979), (Dyson and Farrell 1985). Initial genetic dissections of EBV identified one viral protein, Epstein-Barr Nuclear Antigen 1 (EBNA1), and one region of the viral genome, termed oriP, as being necessary and sufficient for replication of the viral plasmid (Yates, Warren et al. 1984)(Lupton and Levine 1985; Reisman, Yates et al. 1985), (Reisman and Sugden 1986). oriP is composed of two separable cis elements, the Family of Repeats (FR) and the Dyad Symmetry element (DS) (Baer, Bankier et al. 1984; Reisman, Yates et al. 1985), which both contain binding sites for EBNA1 in vitro (Rawlins, Milman et al. 1985)(Harrison, Fisenne et al. 1994), and in vivo (Hsieh, Camiolo et al. 1993), (Niller, Glaser et al. 1995) (Figure 1). FR is composed of 21 imperfect copies of a 30bp repeat and contains 20 high affinity EBNA1-binding sites (Figure 1). When FR is bound by EBNA1, it both serves as a transcriptional enhancer of promoters in cis up to 10kb away (Reisman and Sugden 1986), (Yates 1988), (Sugden and Warren 1989), (Wysokenski and Yates 1989), (Gahn and Sugden 1995), (Kennedy and Sugden 2003; Altmann, Pich et al. 2006), and contributes to the nuclear retention and faithful maintenance of FR-containing plasmids (Langle-Rouault, Patzel et al. 1998), (Kirchmaier and Sugden 1995), (Wang, Lindner et al. 2006) (Nanbo, Sugden et al. submitted). The efficient partitioning of oriP plasmids is also likely attributable to FR. Single cell experiments in which oriP plasmids are visualized in live cells have demonstrated that at least 88% of those plasmids that are duplicated are co-localized with one another in S-phase and are partitioned faithfully to each daughter cell (Nanbo, Sugden et al. submitted). While the virus has evolved to maintain 20 EBNA1-binding sites in FR, efficient plasmid maintenance requires only 7 of these sites, and can be reconstituted by a polymer of 3 copies of DS, having a total of 12 EBNA1-binding sites (Wysokenski and Yates 1989). The Dyad Symmetry element (DS) is sufficient for initiation of DNA synthesis in the presence of EBNA1 (Aiyar, Tyree et al. 1998), (Yates, Camiolo et al. 2000), and initiation occurs either at or near DS (Gahn and Schildkraut 1989); (Niller, Glaser et al. 1995). Termination of viral DNA synthesis is thought to occur at FR, because when FR is bound by EBNA1 it functions as a replication fork barrier as observed by 2D gel electrophoresis (Gahn and Schildkraut 1989), (Ermakova, Frappier et al. 1996), (Wang, Lindner et al. 2006). Initiation of DNA synthesis from DS is licensed to once-per-cell-cycle (Adams 1987), (Yates and Guan 1991), and is regulated by the components of the cellular replication system (Chaudhuri, Xu et al. 2001), (Ritzi, Tillack et al. 2003), (Dhar, Yoshida et al. 2001), (Schepers, Ritzi et al. 2001), (Zhou, Chau et al. 2005), (Julien, Polonskaya et al. 2004). Single cell analyses have shown that DS is a highly-efficient origin; at least 84% of plasmids carrying oriP in cis in HeLa cells expressing EBNA1 are duplicated in each S-phase (Nanbo, Sugden et al. submitted). DS contains four EBNA1-binding sites, albeit with lower affinity than those found in FR (Reisman, Yates et al. 1985). The topology of DS is such that the four binding sites are arranged as two pairs of sites, with 21bp center-to-center spacing between each pair and 33bp center-to-center spacing between the two non-paired internal binding sites (Figure 1.C) (Baer, Bankier et al. 1984; Rawlins, Milman et al. 1985). As there are 10.5bp per turn of B-form DNA, this arrangement positions two EBNA1 dimers on the same helical face of the DNA when bound to a pair of sites. The outer binding sites of each pair (Sites 1 and 4) are of higher affinity than the two internal binding sites (Sites 2 and 3), and several groups have demonstrated by an electrophoretic mobility shift assay (EMSA) that the apparent affinities of EBNA1 for Sites 2 and 3 are increased when their paired high-affinity sites are also occupied. (Harrison, Fisenne et al. 1994), (Summers, Barwell et al. 1996), (Bashaw and Yates 2001). Upon binding to sites in DS, EBNA1 bends the DNA and also causes a permanganate-accessible distortion in the outer binding sites (Harrison 1994, (Frappier and O'Donnell 1992), (Hsieh, Camiolo et al. 1993), (Bashaw and Yates 2001), (Bochkarev, Barwell et al. 1996). Additionally, there is a region of helically unstable DNA near DS that has been proposed to facilitate DNA unwinding and the initiation of DNA synthesis originating at DS (Polonskaya, Benham et al. 2004). Flanking the pairs of EBNA1-binding sites of DS are three 9bp elements that resemble telomeric repeats, and have been termed nonamers (Niller, Glaser et al. 1995), (Yates, Camiolo et al. 2000). Lieberman and colleagues have shown in several studies that these elements are bound by cellular proteins associated with the telomeres, including TTAGGG-Repeat binding Factors (TRF1, TRF2), the human repressor activator protein 1 (hRap1), and tankyrase (Deng, Lezina et al. 2002), (Deng, Atanasiu et al. 2003), (Deng, Atanasiu et al. 2005). Deletion or mutation of the nonamer elements decreases by twofold the average number of plasmids maintained per cell (Deng, Lezina et al. 2002), (Deng, Atanasiu et al. 2003). One protein found to bind these elements, TRF2, exhibits cooperative binding with a truncated form of EBNA1 at DS as measured in gel shift assays, and may contribute to the enhanced replication of DNA by increasing the apparent affinity of EBNA1 for its binding sites (Deng, Lezina et al. 2002). Despite this two-fold effect, the nonamer elements are not necessary for replicative function (Julien, Polonskaya et al. 2004), (Yates, Camiolo et al. 2000), (Koons, Van Scoy et al. 2001), (Deng, Lezina et al. 2002), (Deng, Atanasiu et al. 2003; Wang, Lindner et al. 2006). Genetic dissections of DS indicate that the minimal requirement for its replicative function is one pair of EBNA1-binding sites (Harrison, Fisenne et al. 1994), (Shirakata and Hirai 1998), (Yates, Camiolo et al. 2000), (Koons, Van Scoy et al. 2001), (Bashaw and Yates 2001). The spacing between a pair of EBNA1-binding sites is critical as well; changing the center-to-center spacing of the sites by inserting or deleting only 1bp eliminates DNA synthesis (Bashaw and Yates 2001). Insertions of 3, 5, or 10bp between the paired sites do not support DNA replication either (Harrison, Fisenne et al. 1994). Additionally, pairs of EBNA1-binding sites within the EBV genome that have center-to-center spacings other than 21bp, such as in FR (with 30bp spacing) and near the Q promoter (with 24bp spacing) (Figure 1.A), do not support DNA replication (Reisman, Yates et al. 1985), (Aiyar, Tyree et al. 1998), (Yates, Camiolo et al. 2000). The functional roles of the elements within DS have been confirmed by studies of another region of EBV’s genome, termed Rep*, which was identified as an element that can substitute for DS inefficiently (Kirchmaier and Sugden 1998). Polymerizing Rep* eight times yielded an element as efficient as DS in its support of replication (Wang, Lindner et al. 2006). Biochemical dissection of Rep* identified a pair of EBNA1-binding sites with a 21bp center-to-center spacing critical for its replicative function (ibid). These sites are also bent when bound by EBNA1 (ibid). The minimal replicator of Rep* was found to be the pair of EBNA1-binding sites, as replicative function was retained even after all flanking sequences in the polymer were replaced with sequences derived from lambda phage which were known not to support DNA synthesis themselves (ibid). These findings indicate that neither the flanking nonamer elements nor the helically unstable regions are essential for initiation of DNA synthesis. Comparisons of DS and Rep* have revealed a common mechanism: these replicators support the initiation of DNA synthesis by recruiting the cellular replicative machinery via a pair of appropriately spaced sites, bent and bound by EBNA1.

Figure 1.

The EBV genome and the latent origin of plasmid replication (oriP). (A) The genome of EBV is depicted here in its covalently closed circular form of ~165kbp. The original genomic fragments produced by BamHI digestion are labeled with letters on the interior. Promoters, transcription units, and gene products of the latent cycle are labeled on the exterior as arrows, dashed lines, and open boxes, respectively. In bold boxes are the lytic origin of DNA synthesis (oriLyt), the latent origin of DNA synthesis (oriP), the terminal repeats (TR), and the ~14kbp region shown to support initiation of DNA synthesis in the Raji strain of EBV (Raji ori). (B) The origin of plasmid replication (oriP) is composed of the Family of Repeats (FR) and the Dyad Symmetry (DS) element. The Rep* element is located 240bp downstream of DS, and the original mapping of this element is shown as a lightly shaded box. EBNA1-binding sites are depicted as open boxes in all three cis elements; the dyad present in DS is denoted by the head-to-head arrows above it. The base positions of each of these elements are listed below them with respect to the sequence of the B95–8 strain of EBV. (C) A more detailed depiction of the DS element. The two pairs of EBNA1-binding sites of DS have an intra-pair spacing of 21bp from the center of one site to the center of the paired site, and an inter-pair spacing of 33bp between the centers of the adjacent binding sites. Nine base pair binding sites for the TTAGGG-repeat Binding Factor 2 (TRF2), also referred to as 'nonamers,' flank the pairs of EBNA1-binding sites and are denoted by the shaded boxes. The arrows above the TRF2-binding sites denote the N-to-C terminal orientation that TRF2 would use to bind to these sites.

Not all origins of DNA synthesis in the EBV genome are discrete, though. In the Raji strain of EBV, a 14kbp region is used predominantly to initiate licensed DNA synthesis (Adams 1987), (Little and Schildkraut 1995), (Norio, Schildkraut et al. 2000), (Norio and Schildkraut 2004). The initiation of DNA synthesis is spread across this large region, and is reminiscent of the broad zones of initiation found with many chromosomal origins (reviewed in (Bogan, Natale et al. 2000)). However, the mechanism of the initiation of DNA synthesis from Raji ori is unlike that of DS and Rep*; EBNA1 does not detectably bind to it and no single origin has been localized within it that would facilitate its analysis (C. Wang and Sugden unpublished results).

3. EBNA1: The Sole Trans-Acting Element of EBV Required for Its Plasmid Replicon

Epstein-Barr Nuclear Antigen 1 (EBNA1) is the viral initiator protein of EBV (Lupton and Levine 1985; Yates, Warren et al. 1985), because binding of EBNA1 to a pair of sites in DS or Rep* converts that DNA into an origin of DNA synthesis (Gahn and Schildkraut 1989), (Wang, Lindner et al. 2006). The 641 amino acids (AA) of EBNA1 have been categorized into domains associated with its varied functions by mutational and deletional analyses (Figure 2). Two regions, between AA40-89 and AA329–378 are capable of linking two DNA elements in cis or in trans when bound by EBNA1, and have thus been termed Linking Region 1 and 2 (LR1, LR2) (Middleton and Sugden 1992), (Frappier and O'Donnell 1991), (Su, Middleton et al. 1991), (Mackey, Middleton et al. 1995). Fusing these domains of EBNA1 to GFP homes the GFP to mitotic chromosomes (Marechal, Dehee et al. 1999), (Kanda, Otter et al. 2001). LR1 and LR2 are functionally redundant for replication; a deletion of either one yields a derivative of EBNA1 capable of supporting DNA replication (Mackey and Sugden 1999), (Sears, Ujihara et al. 2004). LR1 and LR2 are rich in arginine and glycine residues, and resemble the AT-hook motifs that bind A/T rich DNA (Aravind and Landsman 1998), (Sears, Ujihara et al. 2004). An in vitro analysis of LR1 and LR2 of EBNA1 has demonstrated their ability to bind to A/T rich DNA (Sears, Ujihara et al. 2004). When LR1, containing one such AT-hook, was fused to the DNA-binding and dimerization domain of EBNA1, it was found to be sufficient for DNA replication of oriP plasmids, albeit less efficiently than the wild-type EBNA1 (ibid). LR1 and LR2 do differ, though. The C-terminal half of LR1 is composed of amino acids other than the repeated Arg-Gly of the N-terminal half, and is termed Unique Region 1 (UR1). UR1 is necessary for EBNA1 to activate transcription efficiently from transfected and integrated reporter DNAs containing FR (Wu, Kapoor et al. 2002; Kennedy and Sugden 2003; Altmann, Pich et al. 2006). UR1 is also essential for the efficient transformation of B-cells infected by EBV. When a derivative of EBNA1 lacking this domain replaces the wild-type protein in the context of the whole virus, these derivative viruses have 0.1% of the transforming ability of the wild-type virus (Altmann, Pich et al. 2006). The presence of the UR1 domain in EBNA1 allows a shift from the Wp promoter, which is used immediately after infection, to the Cp promoter 5 days post-infection (Figure 1.A) (ibid). This shift in promoter usage correlates with the transcription of several viral genes and a prevention of the infected cell from dying by apoptosis, events likely underlying the role of transcriptional activation by UR1 in transforming infected B-cells. Additionally, UR1 is necessary for the continuing survival of EBV-positive B-cells; derivatives of EBNA1 lacking UR1 function as dominant-negative mutants and inhibit cell survival (Kennedy, Komano et al. 2003).

Figure 2.

A Domain-Based Model and Partial Structural Representation of Epstein-Barr Nuclear Antigen 1 (EBNA1). EBNA1 is depicted as modules corresponding to functional domains identified by mutational or deletional analyses. Amino acid (AA) positions are denoted below the junction of adjacent domains. G-R denotes a glycine-arginine rich region within the linking regions 1 (AA40–89) and 2 (AA325–379). Two regions composed of more unique amino acids within LR1 and LR2 are denoted as Unique Region 1 (UR-1) and Unique Region 2 (UR-2), respectively. G-G-A denotes a glycine-glycine-alanine region that composes over a third of the molecule. NLS denotes a nuclear localization signal. The DNA-binding and dimerization domain is labeled according to its functions, and has been crystallized and its structure determined by x-ray diffraction. The acidic tail is rich in aspartic and glutamic acids. Functions attributed to EBNA1 are denoted in rectangles below the modular depiction of EBNA1; additional proteins shown to interact with specific regions of EBNA1 are denoted in rounded rectangles above it.

A yeast two-hybrid assay found that LR2 of EBNA1 associates with the cellular protein EBP2, an essential nucleolar protein implicated in ribosome biogenesis (Shire, Ceccarelli et al. 1999). The maintenance of oriP plasmids in S. cerevisiae is made possible when human EBP2 is co-expressed with EBNA1 (Kapoor, Shire et al. 2001), (Kapoor and Frappier 2003). These findings in yeast have not been extended to human cells, though. EBP2 was found to bind LR2 of EBNA1, and LR2 is not required for EBNA1’s support of oriP replication (Shire, Ceccarelli et al. 1999), (Mackey and Sugden 1999), (Sears, Ujihara et al. 2004). Additionally, the N-terminal half of EBNA1 can be replaced with cellular proteins containing AT-hook motifs, such as HMGA1a, and still retain replicative function (Hung, Kang et al. 2001), (Sears, Kolman et al. 2003) (Altmann, Pich et al. 2006). These findings indicate that it likely is the AT-hook activities of LR1 and LR2 and not LR2’s binding of EBP2 that are required for the maintenance of oriP in human cells.

A third of EBNA1's residues (AA91–328) consist of glycine-glycine-alanine (GGA) repeats, implicated in EBNA1’s ability to evade the host immune response by inhibiting proteosomal degradation and presentation (Levitskaya, Coram et al. 1995), (Levitskaya, Sharipo et al. 1997). These repeats have also been found to inhibit translation of EBNA1 in vitro and in vivo (Yin, Manoury et al. 2003). However, the deletion of much of this domain has no apparent effect on functions of EBNA1 in cell culture, making the role that this domain plays difficult to elucidate.

A nuclear localization signal (NLS) is encoded by AA379–386, which also associates with the cellular nuclear importation machinery (Kim, Maher et al. 1997), (Fischer, Kremmer et al. 1997). Sequences within the Arg-Gly rich regions of LR1 and LR2 may also function as NLSs due to their highly basic content. The ubiquitin specific protease 7 has been found to bind to a region of EBNA1 downstream of its identified NLS (Saridakis, Sheng et al. 2005).

Lastly, the C-terminus (AA458–607) encodes the overlapping DNA-binding and dimerization domains of EBNA1. The structure of these domains bound to DNA has been solved by x-ray crystallography, and was found to be similar to the DNA-binding domain of the E2 protein of papillomaviruses (Hegde, Grossman et al. 1992), (Kim, Tam et al. 2000), (Bochkarev, Barwell et al. 1996). Each monomer of this domain of EBNA1 is a mirror-image of its partner, with the two-fold axis of the dimer aligning with the two-fold axis of its 16bp binding site (ibid). The dimerization domain is formed by an eight-stranded, anti-parallel, beta barrel, with four strands being contributed by each monomer. The residues that contribute to sequence-specific DNA binding lie predominantly between AA458–477 as an extended chain that threads through the minor grooves of each half of the binding site, as well as the N-terminal cap of an alpha helix (ibid). An additional alpha helix (AA514–527) with structural similarity to that used by papillomaviral E2 proteins to contribute to their DNA-binding has also been implicated in site-specific binding (Bochkarev, Barwell et al. 1996), (Cruickshank, Shire et al. 2000). A symmetrical underwinding and compensatory overwinding of the DNA helix is present when bound by EBNA1, as well as a subtle bending of the DNA toward the protein (Bochkarev, Barwell et al. 1996). However, the extent of bending is not as pronounced as was observed in biochemical assays in which the DNA and protein were in solution (Bashaw and Yates 2001), (Wang, Lindner et al. 2006). In silico modeling of a pair of EBNA1-binding sites with 21bp center-to-center spacing with the degree of bending observed in the co-crystal structure (~10°) predicted steric hindrance between the two EBNA1 dimers resulting from significant overlap of the calculated electron densities (Bochkarev, Barwell et al. 1996). The measured bending by EBNA1 of two sites is on the order of 45° which indicates that the exact structure of EBNA1 bound to DNA is uncertain (Bashaw and Yates 2001), (Wang, Lindner et al. 2006).

Assays measuring EBNA1’s binding to various DNA sequences have demonstrated that it recognizes many related 16-mers, making in silico prediction of its binding sites difficult (Ambinder, Shah et al. 1990), (Wang, Lindner et al. 2006). Each monomer of EBNA1 contacts portions of DNA extending over eight base pairs and although many 16-mers known to bind EBNA1 dimers are palindromes, they need not be (Baer, Bankier et al. 1984), (Rawlins, Milman et al. 1985), (Wang, Lindner et al. 2006). This complexity has also made it difficult to predict those sites bound by EBNA1 under physiologic conditions. All known transcriptional and replicative functions of EBNA1 depend on its binding of DNA site-specifically. For example, a derivative of EBNA1 that is capable of localizing to the nucleus and binding DNA, but lacks all other functions, behaves dominant-negatively for all of EBNA1’s activities (Kirchmaier and Sugden 1997).

4. EBV usurps cellular factors for its own ends

EBNA1 lacks any enzymatic activities despite all functions attributed to it and must rely entirely on cellular factors to carry out these functions (Yates 1988). Many proteins associated with EBNA1’s replicative functions have been identified recently, while the cellular factors associated with its ability to modulate transcription are unknown. EBNA1-dependent DNA synthesis from both DS and Rep* is licensed to occur once-per-cell-cycle, and several groups have shown by chromatin immunoprecipitation (ChIP) that this process is mediated and regulated by the cellular replicative machinery (Schepers, Ritzi et al. 2001), (Dhar, Yoshida et al. 2001), (Ritzi, Tillack et al. 2003), (Chaudhuri, Xu et al. 2001), (Wang, Lindner et al. 2006). When EBNA1 binds to a pair of binding sites of DS or Rep*, the protein-DNA structure formed by this interaction recruits members of the pre-Replicative Complex (pre-RC) to it (ibid). The Origin Recognition Complex (ORC1–6) and the putative replicative helicase (MCM2–7) have both been found to associate with these EBV origins in a similar cell-cycle dependent manner as is thought to occur generally in human cellular origins and been found for the human lamin B2 origin (Chaudhuri, Xu et al. 2001), (Dhar, Yoshida et al. 2001), (Ritzi, Tillack et al. 2003), (Schepers, Ritzi et al. 2001), (Zhou, Chau et al. 2005), (Julien, Polonskaya et al. 2004), (Wang, Lindner et al. 2006), (Abdurashidova, Danailov et al. 2003). The replication of oriP plasmids is significantly inhibited in cells hypomorphic for ORC2 or overexpressing the cellular protein geminin that inhibits full assembly of the pre-RC, confirming EBV’s reliance on these cellular pathways for the replication of its own genome (Dhar, Yoshida et al. 2001). While a pair of EBNA1-binding sites is sufficient for DNA synthetic function, the presence of the nonamer elements flanking the origin and the binding of TRF2 to them stimulates the initiation of DNA synthesis from that origin (Deng, Lezina et al. 2002; Deng, Atanasiu et al. 2003; Atanasiu, Deng et al. 2006). Some of this stimulation likely results from TRF2’s increasing EBNA1’s apparent affinity for DS; in addition, some may also result from the direct interaction of TRF2 with the BAH domain of ORC1, allowing increased recruitment of ORC1 to DS at the G1/S border of the cell cycle (Ritzi, Tillack et al. 2003; Atanasiu, Deng et al. 2006). The BAH domain of ORC1 is important for the activation of EBNA1-dependent origins of DNA synthesis. Deletion of the entire domain (AA1–169), as well as a single point mutation (E111K), reduced the association of ORC1 with DS to background as measured by Chromatin ImmunoPrecipitation (ChIP) (Noguchi, Vassilev et al. 2006). These same modifications also reduced the short-term replication of oriP plasmids to background levels, and the overexpression of the BAH domain alone (AA 1–315) functioned dominant-negatively to inhibit oriP replication (ibid). Cellular components in addition to ORC used for DNA synthesis, such as polymerases, single-stranded DNA-binding proteins and other accessory proteins must also be recruited to EBNA1-dependent origins of DNA synthesis. EBV does not express viral proteins with these functions during the latent phase of it’s life cycle.

Although herpesviruses do not use histones to coat their genomes when compacted in viral particles, EBV does use cellular histones to condense its 55 microns of duplex DNA into nucleosomes while resident in the host (Wensing and Farrell 2000), (Shaw, Levinger et al. 1979), (Dyson and Farrell 1985). The nucleosomal structure of oriP is defined by EBNA1 binding to it; EBNA1 can displace histones covering its binding sites in vitro and can maintain DS and FR free of nucleosomes when binding to them (Avolio-Hunter, Lewis et al. 2001; Avolio-Hunter and Frappier 2003; Zhou, Chau et al. 2005). The positioning of nucleosomes by EBNA1 on oriP plasmids occurs as quickly as 6 days post-transfection, and the histones flanking DS undergo cell-cycle-dependent remodeling via acetylation and methylation, apparently by using the cellular SNF2 and HDAC2 proteins (Zhou, Chau et al. 2005). The role of this chromatin remodeling at DS on its replicative function is uncertain, though; it occurs both in Raji cells in which DS apparently is not used as the predominant origin of DNA synthesis (Norio, Schildkraut et al. 2000) and in an engineered variant of EBV DNA in which DS is the only origin shown to function (ibid). However, some epigenetic events such as nucleosomal remodeling may be important for the replication of oriP plasmids and by inference for the EBV genome too (Leight and Sugden 2001). The introduction of oriP plasmids into cells is followed by their precipitous loss at a rate of ≥25% of these plasmids per generation from the cell population, despite their being replicated (Leight and Sugden 2001). This phenomenon is observed with plasmids derived from either mammalian or bacterial sources where the associated proteins are removed from the plasmid DNA during its purification (ibid). By 2–3 weeks after their introduction the remaining plasmids (1–10% of those initially introduced) that are duplicated are stably maintained in >99.7% of mitoses and are thus described as being “established” (Leight and Sugden 2001; Nanbo, Sugden et al. submitted). Single-cell analysis of established plasmids in live cells has demonstrated that any subsequent loss of plasmids results from defects in DNA synthesis, which lead to 16% of plasmids failing to duplicate each S-phase (Nanbo, Sugden et al. submitted). These observations indicate a requirement for an epigenetic change to occur for these plasmids to be established, potentially by creating the appropriate nucleosomal architecture near FR and DS to allow efficient DNA synthesis and plasmid maintenance.

5. A Proposed Model for the Sequential Assembly of Cellular Proteins at an EBNA1-Dependent Origin of DNA Synthesis

The study of the replicative mechanisms by which DS functions has been tractable and fruitful because of the several key attributes of DS described above. These experiments elucidating both the cis (short and long-term replication assays) and trans elements (mass spectrometry, chromatin immunoprecipitation) that contribute to the licensed initiation of DNA synthesis at DS have led us to propose the following model (Figure 3). A pair of EBNA1-binding sites, spaced precisely 21bp center-to-center, are bound and bent in the same direction by two dimers of EBNA1. EBNA1 binds to a pair of sites in DS interacting cooperatively with the TTAGGG-repeat Binding Factor 2 (TRF2), which binds to an adjacent 9bp element termed a ‘nonamer.’ EBNA1 and TRF2 associate with DS throughout the cell cycle, as do subunits 2 through 6 of the Origin Recognition Complex (ORC2–6) (Figure 3.A). ORC1 associates with DS in G1/S, as it does with cellular origins. The recruitment of ORC1 to DS does not require TRF2-binding sites. However, the association of the N-terminal basic domain of TRF2 with the N-terminal BAH domain of ORC1 enhances the association of ORC1 with DS. The regulatory protein Cdc6 is also likely recruited to DS in G1/S (Figure 3.B). ORC1–6 and Cdc6 function as a clamp loader for the reiterative recruitment and loading of the putative replicative helicase MCM2–7 to yield a head-to-head double hexamer at DS. The association of MCM2–7 with DS is dependent upon the regulatory protein Cdt1, which possibly associates with DS in complex with MCM2–7 (Figure 3.C). Taken together, this protein/DNA complex progresses from a pre-Replicative Complex (pre-RC) to a preInitiation Complex (pre-IC), allowing the efficient initiation of DNA synthesis to occur at or near DS in at least 84% of S-phases.

Figure 3.

A sequential model of the licensing process for DNA synthesis at an EBNA1-dependent origin. (A) A pair of EBNA1 dimers binds to a pair of 16bp binding sites with a precise 21bp center-to-center spacing, placing the dimers on the same helical face of the DNA. The binding event bends the DNA toward the DNA-binding and dimerization domain in an additive manner and in the same direction. The TTAGGG-repeat Binding Factor 2 (TRF2) is found as a dimer in the cell, and one monomer can bind to a 9bp element that flanks the pairs of EBNA1-binding sites in wtDS. The binding of TRF2 to these elements (also termed nonamers) permits the cooperative binding of EBNA1 to its binding sites, thereby increasing EBNA1’s apparent affinity. In addition, ORC2–6 were found to be bound to such an EBNA1-dependent origin throughout the cell cycle by Chromatin Immunoprecipitation (ChIP). (B) ORC1 has been shown to interact with EBNA1-dependent origins in the presence and absence of the TRF2 binding sites, although ORC1’s association with these origins is likely enhanced by the interaction of it’s BAH domain with the N-terminal basic domain of TRF2. Cdc6 is also likely recruited to the origin in G1/S. (C) Taken together, these assembled proteins at the EBNA1-dependent origin allow the subsequent recruitment of, and likely the reiterative loading of, the putative replicative helicase MCM2–7 as a head-to-head double hexamer, possibly in a complex with the regulatory protein Cdt1.

6. Conclusions

EBV infects small, dense B-lymphocytes which are quiescent, induces them to enter S-phase, and maintains them in a proliferating state in cell culture. EBV introduces its DNA into these resting cells covered only with polyamines and must insure that its replicon is handled by its host cell as it transits from G0 to G1 to insure its being recognized and used by the cellular replicative machinery in the first S-phase. These requirements appear to be met by EBNA1 binding to and bending DS, along with the positioning of nucleosomes adjacent to it. This bent protein/DNA structure, which is aided by cellular proteins such as TRF2 binding adjacent to it, efficiently recruits the cellular replicative machinery for the duplication of the viral genome. Moreover, this EBNA1-dependent origin functions under the same regulatory constraints imposed upon cellular origins of DNA synthesis. As there are no known consensus sequences that define an origin of DNA synthesis in human cells, an overall protein/DNA structure may be the predominant identifying mark of an origin, which EBV has mimicked by using it own cis and trans elements.