Initiation of lytic DNA replication in Epstein–Barr virus: search for a common family mechanism

Abstract

Herpesviruses are a complex family of dsDNA viruses that are a major cause of human disease. All family members share highly related viral replication proteins, such as DNA polymerase, ssDNA-binding proteins and processivity factors. Consequently, it is generally thought that lytic replication occurs through a common and conserved mechanism. However, considerable evidence indicates that proteins controlling initiation of DNA replication vary greatly among the herepesvirus subfamilies. In this article, we focus on some of the known mechanisms that regulate Epstein-Barr virus lytic-cycle replication, and compare this to other herpesvirus family members. Our reading of the literature leads us to conclude that diverse viral mechanisms generate a common nucleoprotein prereplication structure that can be recognized by a highly conserved family of viral replication enzymes.

Epstein–Barr virus (EBV) is the prototypical γ-herpesvirus because of its early discovery as a human tumor virus [1]. EBV is now known to contribute to a variety of human disorders, including infectious mononucleosis, oral hairy leukoplakia, nasopharyngeal carcinoma, Burkitt’s lymphoma and lymphoproliferative diseases occurring in immunocompromised individuals. A second human γ-herpesvirus has been identified as the causative agent of Kaposi’s sarcoma, and has been designated Kaposi’s sarcoma-associated herpes virus (KSHV) [2]. KSHV, a close relative of EBV, is also found to be associated with lymphoid disorders, including pleural effusion lymphomas and Castlemen’s disease. Both EBV and KSHV preferentially establish latent infection in B lymphocytes and contribute to malignant transformations in both lymphoid and epithelial tissues.

Although latent viral infection is typically thought to contribute to lymphomagenesis, it is becoming increasingly apparent that lytic replication of the virus is a strong contributing factor to cancer cell evolution. Lytic EBV has been implicated in nasopharyngeal carcinoma [3] and methotrexate-induced lymphomas, arising in patients treated for rheumatoid arthritis and polymyositis [4]. Chronic lytic EBV infection caused by co-infection of malaria is suspected of promoting endemic Burkitt’s lymphoma [5,6]. KSHV lytic infection in endothelial cells is strongly correlated with progression of Kaposi’s sarcoma. Pharmacological inhibitors of herpesvirus lytic replication can ameliorate disease progression, but do not prevent recurrence due to drug resistance [7]. Considering the importance of lytic-cycle replication to pathogenesis, a further understanding of the early events that control initiation of viral DNA replication will improve our ability to develop therapeutics of viral-associated disease.

Initiation of γ-herpesvirus lytic cycle

Like all herpesviruses, lytic replication can initiate at two points in the viral lifecycle: during primary infection or upon reactivation of latent infection. It is unclear whether these mechanisms are significantly different, but it is likely that different chromosome configurations and cell-response factors must be involved in these different pathways to productive infection. For EBV and KSHV, the immediate-early (IE) proteins must be expressed and functional for productive infection to progress. While EBV and KSHV share partial conservation of these IE proteins, they have remarkably different biological and biochemical properties. The most conserved IE protein is referred to as Rta, and is essential for transcription activation and lytic replication in both viruses [8,9]. In EBV, a second IE protein, referred to as Zta (encoded by the BZLF1 gene and also known as, Z, ZEBRA and EB1) plays a primary role in lytic activation and lytic replication [10–13]. The KSHV ortholog, K8, does not appear to activate transcription [8,14] and overexpression of BRLF1, the gene that encodes Rta, can overcome the block to lytic replication in a K8-null virus [15]. This divergence in Zta/K8 requirement emphasizes the variation in mechanisms of initiation of lytic replication, even among two highly related γ-herpesviruses.

The requirement for Zta and Rta in EBV lytic replication have been demonstrated by numerous genetic and biochemical studies. Genetic disruption of either gene prevents lytic replication [9]. Although viruses lacking Zta can still immortalize primary B lymphocytes in culture, these cells fail to form tumors in severe-combined immunodeficient mice [16,17]. Since Zta is a potent transcription activator, it is likely that Zta expression promotes tumor formation through activation of viral and cellular factors, including viral cytokines that promote tumor formation [18–20]. Rta and Zta can be coexpressed from a single bicistronic transcript in EBV [21], and a similar gene organization exists for KSHV ORF50, the gene that encodes Rta, and the K8 gene [22]. Signaling pathways that activate Rta or Zta transcription are known to initiate lytic-cycle gene expression. Numerous cellular factors can bind the transcriptional regulatory regions of these IE genes and are subject to complex regulation. In addition, both Rta and Zta can interact with numerous cellular factors and are subject to post-translational modifications that can affect their function in lytic replication and transcription activation. Thus, regulation of these IE genes represents an important level of control for initiation of lytic-cycle replication. Many of these controls have been reviewed else-where [23–25]. In this article, we focus on the role of these and other proteins in the establishment of an active origin of lytic replication, and consider their function in this later stage process where they have essential and direct functions at the origins of DNA replication.

Identification of the conserved Herpesviridae core replication machinery

In 1986, a landmark complementation assay was developed to identify six core herpes simplex virus (HSV)1 genes required to support lytic replication of a plasmid containing the HSV1 repeated short-region origin of replication (OriS) [26]. Subsequent studies revealed that orthologs of these six replication proteins were required for lytic replication of other herpesviruses, including EBV [27–30]. In the case of EBV, these include a viral polymerase (BALF5), a polymerase processivity factor (BMRF1/EA-D), a helicase–primase complex (BBLF4, BSLF1 and BBLF1/2) and a ssDNA-binding protein (BALF2) [27]. The BALF5 polymerase is highly processive, able to efficiently add more than 7200 nucleotides to an RNA primer on a DNA template at a rate of 12 nucleotides per second before falling off and its affinity for RNA–DNA hybrids is greater than its affinity for ssDNA [31]. BALF5 protein has 3′–5′ exonuclease activity and preferentially excises a terminal mismatched nucleotide [32]. Affinity analysis has shown that BALF5 polymerase is able to interact with the helicase/primase proteins BSLF1, BBLF4 and BBLF2/3 [33], which themselves are able to form an enzymatic complex independently [34]. The processivity power of BALF5 is enhanced by its interaction with the BMRF1 gene product, EA-D [35,36]. EA-D, a sliding clamp protein resembling the cellular PCNA protein, is required for lytic replication [37] and has a unique role as a transcription factor [38,39]. These herpesvirus DNA replication enzymes bear high sequence homology across family members, are capable of both leading and lagging strand synthesis in vitro [40] and can synthesize DNA at a preformed replication fork in vitro [41]. Furthermore, many of these core replication proteins can be interchanged between different herpresviruses; HSV1 core replication proteins replicate the varicella zoster virus genome [42]; the EBV core proteins replicate human cytomegalovirus (CMV) [43] and herpesvirus papio DNA [44]; and the KSHV core proteins replicate the EBV genome [30]. The herpesvirus core replication machinery can even replicate DNA viruses from other viral families, such as adeno-associated virus [45,46] and simian virus 40 [47]. However, each viral lytic origin requires specific recognition conferred by its own origin-binding protein. Without the correct origin-binding protein, the core enzymatic complex cannot initiate replication.

Diverse group of origin-binding proteins

A herpes lytic origin-binding activity was first described for HSV1 [48] and later identified as the product of the UL9 gene [49]. UL9 encodes an ATP-dependent helicase [50,51], which, in cooperation with the ssDNA-binding protein ICP8, is able to bind, loop, distort and unwind OriS DNA [52–56]. Human CMV uses a UTPase encoded by the UL84 gene as its origin-binding protein [43,57,58]. UL84 protein binds an RNA stem-loop structure within the CMV origin of lytic replication (OriLyt) [59] where it functions as a transcription factor along with the IE2 protein at the origin’s bidirectional promoters [60]. In the case of EBV, the IE transcription factor Zta is the best candidate for a viral-encoded origin-binding protein required for lytic replication [27,61–63].

Zta binds directly to multiple sites in OriLyt and recruits components of the viral core replication machinery to OriLyt [64–67]. In addition to OriLyt binding, Zta activates transcription from multiple viral promoters, including the two divergent promoters within the approximately 1 kb OriLyt. Zta consists of an amino-terminal transcription activation domain and carboxy-terminal basic-zipper (bZip) domain (Figure 1). The bZip domain has closest homology to the cellular transcription factors C/EBP-α, c-Fos and c-Jun [68,69]. The basic region (residues 170–198) permits binding to origin and promoter DNA via pseudo-palindromic heptad Zta response elements (ZREs), which include AP1/TRE sites [70–72] and others conforming to the loose consensus sequence 5′-T(G>T>C) (A>G>T)(C/G)(T>C>A)(C>A>G)A-3′ found in the promoters of both cellular [72–79] and viral [69,72,80–83] genes. The crystal structure of the bZip domain has revealed a unique fold-back structure at the C-terminus that distinguishes it from all other known bZip structures [84–86]. Recent studies reveal that Zta binds preferentially to methylated cytosines, and this has been proposed to regulate the tendency of EBV to establish latent infection and to efficiently reactivate latent genomes that have been subject to cytosine methylation [83,87–89].

Figure 1. Domain structure of the herpesvirus lytic origin-binding proteins.

Although herpesvirus lytic origin-binding proteins are quite diverse, there are a few limited similarities. HSV UL9 and CMV UL84 share DExD/H box similarities, and most of these proteins contain leucine zipper or pseudo zipper domains (blue and purple boxes). Many of these regions bind to the cellular protein C/EBP-α (also shown).

CMV: Cytomegalovirus; EBV: Epstein–Barr virus; HSV: Herpes simplex virus; KSHV: Kaposi’s sarcoma-associated herpesvirus.

Since Zta lacks any known enzymatic activity, it is thought that cellular-interacting proteins contribute essential activities to replication initiation. Among the many interacting proteins, the interaction with C/EBP-α is most notable because of its potential common role at other herpesvirus lytic origins [90,91] and because it shares the most extensive homology with Zta (Figure 2A). C/EBP-α has been implicated in both Zta-mediated replication and cell-cycle arrest [92–95]. Zta physically interacts with C/EBP-α through contacts in the conserved zipper domain, where the two proteins are thought to form tetramers [96]. The Zta–C/EBP-α interaction is responsible for the induction of CDK inhibitors (p21WAF-1/CIP-1 and p27KIP-1) [95], and in the absence of C/EPB-α, Zta is unable to activate the p21 or p27 promoters or cause cell-cycle arrest [92]. The precise role of cell-cycle arrest prior to viral DNA replication is not completely understood, but is thought to be important for preventing competition of virus with the host cell demands during DNA replication. Cell-cycle arrest prior to DNA replication appears to be a common event for all of the herpesviruses [97].

Figure 2. Zta basic-zipper alignments.

(A) Zta most closely resembles the C/EBP-α protein, the greatest homology is in Zta’s C-terminal tail that includes its critical basic-zipper region. (B) The critical basic region of Zta shares limited homology with the region of HSV1 UL9 protein involved in ssDNA binding.

Like EBV, the KSHV genome encodes a protein, K8 (K-bZip, RAP), which resembles and is syngenic to Zta. Like Zta, K8 binds to C/EPB-α and causes C/EBP-α-dependent G₀/G₁ cell-cycle arrest [98]. However, the requirement for K8 in DNA replication is somewhat controversial. Unlike Zta, ectopic expression of K8 cannot reactivate virus from latency [14]. Furthermore, K8 does not possess any intrinsic DNA-binding activity. Initially known as KSHV RAP, K8 associates with KSHV OriLyt and recruits some viral and cellular proteins to the site of DNA replication [99,100], but its DNA binding is indirect, mediated by interaction with C/EBP-α [101,102], Rta [103] or the viral latency-associated nuclear antigen (LANA) [104]. Initial studies using OriLyt-containing plasmids indicated that K8 is required for lytic replication [105,106], and a K8-knockout (BAC36ΔK8) virus is compromised for DNA replication [8]. However, this defect can be rescued by overexpression of Rta [15], suggesting that, to some extent, K8 is dispensable. By contrast, the overexpression of EBV Rta is unable to rescue a BZLF1-knockout virus [107]. K8 also differs from Zta in that its ‘zipper’ region (residues 190–237), while required for multimerization [108], folds into a β-sheet rather than as an α-helix [109]. K8 can bind to Rta via its zipper motif and this interaction may attenuate Rta transcription activity and, consequently, activate its replication function [103,110]. Thus, K8 has diverged significantly from its EBV ortholog Zta, while KSHV Rta takes on a more prominent role in replication initiation and viral reactivation from latency [8,111,112].

Identification of lytic origins

Early studies with HSV revealed that DNA replication could initiate at three homologous cis-acting elements (two copies of OriS and one copy of the unique long origin of lytic replication [OriL]), each capable of functioning as an origin (Figure 3) [113–116]. Each α-herpesvirus origin (and those of roseolavirus β-herpesviruses) includes a necessary palindrome containing two initiator protein-binding sites (box I and II) flanking an AT-rich sequence [117,118] and a third (box III) adjacent to box I also required for replication in vivo [115,119,120]. EBV typically encodes two homologous copies of OriLyt [121], although there are functional strains that only encode one copy, which includes binding sites for Zta [72,121]. Unlike the origins of α-herpes- and roseolaviruses, the minimal EBV OriLyt sequence is comprised of two essential elements flanking dispensable AT-rich palindromes [122]. These elements are located between two divergent promoters. The upstream essential element (UEE) contains the TATA box for the BHLF1 (or L3) gene, two ZREs (ZRE1/2) [62,122,123], which form an inverted repeat that is similar to the UL9 box III–I binding sites in HSV1 OriS, and a CCAAT box. The downstream essential element (DEE) contains binding sites for the Sp1, Sp3 and ZBP-89 proteins, which interact with the core viral-replication proteins [62,65,124,125], including the EA-D processivity factor that is able to activate the BHRF1 promoter [67] via the downstream element [38]. The DEE also contains a homopurine–homopyrimidine ‘Y-box’ sequence capable of forming a triple helix in vitro, and mutations that impair the triple-helix formation in vitro also disrupt DNA replication in vivo [126]. The BHRL1 promoter also contains two binding sites for Rta, although these are found outside of the minimal OriLyt sequence [127]. The critical regions identified in EBV OriLyt have also been shown to be evolutionarily conserved in the related herpesvirus papio [44].

Figure 3. Human herpesvirus lytic origins of replication.

Inverted repeats containing binding sites for known origin-binding proteins are noted with red boxes. AT-rich regions are boxed in blue and Y-block elements are boxed in green. Flanking divergent genes are shown in yellow. Essential elements have also been noted.

CMV: Cytomegalovirus; DEE: Downstream essential element; EBV: Epstein–Barr virus; EE1: Essential element 1; EE2: Essential element 2; KSHV: Kaposi’s sarcoma-associated herpesvirus; HSV: Herpes simplex virus; IE2RE: Immediate-early gene 2 response element; OriLyt: Origin of lytic replication; OriS: Repeated short-region origin of replication; REE: Rta response element; SRT: Cytomegalovirus OriLyt transcript; UEE: Upstream essential element; ZRE: Zta response element.

Like EBV, KSHV encodes two nearly identical OriLyt sequences found between divergent promoters that are close proximity to GC-rich repeats [111,112,128,129]. These include a region similar to the EBV DEE, containing an Rta responsive element in association with a TATA box, an AT-rich palindrome sequence, and eight CCAAT boxes organized as four head–tail pairs [130] reminiscent of the UL9 (pointed out in [129]) and Zta UEE binding sites on their respective genomes. The closely related murine herpesvirus (MHV)-68 has a set-up similar to KSHV, albeit with two pairs of inverted CCAAT boxes required for efficient replication rather than four [131,132]. Mutation of an OriLyt sequence within MHV-68 not only resulted in the impairment of lytic but also latent replication [133], an interesting observation in light of the recent connection drawn between KSHV LANA and OriLyt [104].

Unlike the other herpesviruses, human CMV encodes only one known OriLyt occupying more than 2.5 kb of the genome [134–139]. This large and complex element, although obviously different from α- and γ-herpesvirus origins, does retain some similarities with the others, including a bidirectional promoter (OriLytPM), many GC-rich inverted repeats, potential transcription-factor binding sites and several AT-rich segments, all of which have currently poorly defined functions [140]. Like EBV, there are two regions of CMV OriLyt known to be essential for replication [60,137]. Essential element 1 (EE1) contains a necessary ‘Y-block’ promoter element [141], similar to that of the EBV OriLyt DEE, which contains binding sites for CMV IE2, UL84 and C/EBP-α [91]. The second essential element (EE2) overlaps a long RNA transcript (CMV OriLyt transcript [SRT]) and contains a base-sensitive RNA–DNA hybrid that can form an RNA stem-loop, to which UL84 also binds [59,142].

Origin strand unwinding & ssDNA binding proteins

Little is known about the early initiation events leading to strand unwinding of γ-herpesvirus lytic origins. The best-characterized lytic origin-binding protein of the family is the HSV1 UL9 helicase, which appears to work in concert with the single-stranded binding protein ICP8 to accomplish strand separation. The functional domains of UL9 have been mapped, including those required for DNA and ICP8 binding [143–146]. In vitro, the UL9 protein binds cooperatively to the OriS box I–III sequence in an ATP-dependent manner [52] and, together with ICP8, is able to unwind several OriS-containing DNA substrates [147,148]. Binding of UL9 and ICP8 causes a conformational change in OriS that is detectable by nuclease probing [149] and electromobility shift assays [150,151]. This activated form of OriS, termed OriS*, has been shown to contain a DNA hairpin formed by complementary intrastrand base pairing of box I and III [152,153]. Nuclear magnetic resonance and DNA melting experiments have been used to demonstrate that a large number of α-herpesviruse have the capacity to form hairpins in OriS at origin-binding protein-recognition sites [154]. It is still unclear as to the role that DNA secondary structure plays in the origins of other herpesviruses, although the presence of so many inverted repeats within their sequences raises the possibility that intrastrand binding may be part of a conserved mechanism of replication initiators.

Although EBV Zta and HSV1 UL9 are structurally two very different proteins, they both possess ssDNA-binding activity [Rennekamp AJ et al., Unpublished Data] [155] and have some sequence homology in the regions known to bind ssDNA (Figure 2B). In addition, the ssDNA-binding proteins encoded by EBV BALF2 and UL29 (ICP8) have approximately 30% sequence homology and high structural equivalency [156–159]. The ICP8 protein is a multifunctional zinc metalloprotein [160], which preferentially binds ssDNA in a nonsequence-specific manner [161]. ICP8 also binds to the C-terminus of the UL9 protein, stimulating helicase activity [55,162–164]. Both ICP8 and EBV BALF2 proteins have the properties of a ssDNA strand annealing protein, similar to the λ RED β recombination protein [165]. ICP8 can displace short DNA strands from their complementary sequences [166], promote DNA strand transfer [167,168] and strand invasion [169–171], and can renature complementary strands of DNA [172]. There is evidence suggesting that the BALF2 protein can perform some, if not all, of these functions as well [173].

θ, rolling circle & recombination?

It has been proposed that herpesviruses copy their genomes via a ‘rolling-circle’ method of lytic replication [174–177]. Several lines of evidence have been used to support this model. First, herpesvirus genomes, including EBV, adopt a circular conformation within the cell quickly following infection and upon lytic induction [178–184]. Circular DNA would provide the template necessary for a rolling-circle mechanism. Second, concatemeric forms of intracellular DNA have been observed as replicative intermediates during lytic replication [121,181,185,186], as well as a reduction in the copy number of genomic termini [187,188]. These intermediates also sediment rapidly in sucrose gradients, demonstrating the presence of viral DNA with molecular weights beyond that of single genomes [189,190]. In addition, some encapsidated defective genomes have been identified as head–tail repeats [191]. Finally, as a proof of principle, it has been demonstrated that the core herpesvirus enzymes and extracts of infected human cells are able to replicate certain templates via a rolling-circle mechanism in vitro [192,193].

However, a simple rolling-circle mechanism does not adequately explain every observation. Further examination of lytic replicative intermediates, using pulse-field gels and electron microscopy, reveals a highly branched network of DNA, containing multiple forks on single molecules [186,194]. In addition, analyses of the kinetics of lytic replication reveal that viral DNA accumulates exponentially and is amplified several hundred-fold in just a few hours [195], while a rolling-circle mechanism would produce a linear amplification [196]. Finally, replicative concatamers of HSV1 genomes contain genomic inversions suggestive of strand-transfer events [181,197]. While these observations do not exclude a rolling-circle mechanism, they suggest that, at the very least, other modes of genome replication are at work. Indeed, EBV OriLyt is able to direct semiconservative replication and production of monomeric progeny soon after lytic induction when incorporated into a plasmid [196]. These observations have led to a dual-mechanism model, similar to that observed in λ-phage replication, where lytic replication is initiated via a plasmid or 'θ’ mode, where copy number is enriched, followed by a switch to rolling circle.

It is also likely that herpesvirus lytic replication involves a recombination mechanism. Herpesvirus genomes are highly recombinogenic, containing frequent genomic inversions anchored by repeat regions [191,198–203]. Homologous recombination at these regions (e.g., the terminal repeats in the EBV genome) occurs frequently and is dependent on lytic replication [168,204–206], and specifically on the core herpes replication machinery and the viral origin (in these experiments, OriS), which together are also sufficient to induce recombination [207,208]. Surprisingly, however, even herpesviruses that naturally lack invertible repeat elements have replication machinery able to support efficient segment inversion, suggesting that the recombinatory function of the conserved core herpes replication machinery plays some additional role, perhaps in replication [209]. The core single-stranded binding protein (e.g., ICP8) is known to promote ssDNA strand invasion, homologous pairing and D-loop formation [168,170,171], not unlike the λRED β recombination protein [165]. In the case of EBV, this protein, encoded by the BALF2 gene, is known to associate with the viral alkaline nuclease (BGLF5) [210], which structurally resembles the λ RED α exonuclease [211]. BGLF5 has 5′–3′ exonuclease activity, as well as endonuclease activity on linear ssDNA, linear dsDNA, nicked dsDNA circles and super-coiled plasmid DNA, in addition to an RNase activity [211–214]. This nuclease is known to contribute to, although not absolutely required for, genome replication [215].

Role for cellular recombination & DNA damage response proteins

In addition to encoding their own proteins capable of promoting recombination, herpesviruses also interface with host cell replication machinery. Many of these proteins are involved in cellular DNA damage repair–recombination pathways [216–218].

Herpes simplex virus 1/2 infection induces phosphorylation of p53, ATM and Mre11–Rad50–Nbs1 (MRN) complex members, as well as several other DNA damage proteins (e.g., RPA, Chk2, Rad50 and 53BP) [219–221]. At the same time, several of these proteins are recruited to viral-replication compartments (e.g., p53, ATM, the MRN complex, DNA–PKCs, Rad50, Ku80/86 and WRN) [219,220,222] and interact with ICP8 [222] and/or OriS [223]. This damage response is not present in latency or latency-like situations, and not inhibited by the viral DNA polymerase inhibitor phosphonoacetic acid [219], suggesting that this is an early replication event. Indeed, RPA, Rad51 and Nbs1 are recruited to prereplicative compartments containing only UL9, ICP8 and the helicase–primase complex in the absence of polymerase.

Generally speaking, activation of the DNA damage response is beneficial for viral replication as Mre11 or ATM (and perhaps WRN) mutant cells have reduced capacity to support lytic replication [219,222]. By contrast, viral replication in Ku70-deficient murine embryonic fibroblasts is increased by almost 50-fold [222], a second subunit of DNA–PK, Ku80/Ku86, is excluded from replication compartments [221], and the DNA–PK core subunit (DNA–PK_CS) is degraded in a proteosome-dependent manner upon expression of HSV1 ICP0 [224]. These observations suggest that the cellular homologous recombination (HR) pathway is important for lytic replication while the nonhomologous end-joining (NHEJ) pathway is inhibitory [221,222]. ICP0 is also capable of inducing the phosphorylation/activation of Chk2 via ATM [225], and knocking out the DNA damage pathway induced by ssDNA and mediated by ATR/ATRIP/RPA [226].

In terms of the DNA damage response, CMV infection appears to be very different from HSV1/2. p53 is increased, phosphorylated and relocated to viral replication compartments [227], and p53-null fibroblasts, while permissive for CMV infection, show a decrease in viral DNA and particle production [228]. However, ATM is not activated and the MRN complex is excluded from replication compartments. Although the quantity and phosphorylation of Nbs1 does increase, both ATM and Mre11 are dispensable for CMV replication [227].

Epstein–Barr virus lytic replication resembles HSV1/2 in this regard; EBV elicits ATM signal transduction (with minimal activation of ATR) and recruits phosphorylated p53, ATM and the MRN complex to replication compartments. However, ATM activation was not required, as caffeine treatment, which inhibits ATM activation, did not affect lytic replication [229]. As in the case of HSV1/2, proteins involved in HR, including RPA, Rad51, Rad52 and the MRN complex, are recruited and loaded onto the EBV genome in replication compartments. Furthermore, EBV replication compartments contained dsDNA breaks and Rad51 and RPA32 were required for viral DNA synthesis [230]. Zta binds to RPA subunits [231] and forms a functional interaction with 53BP1 [232], which is also involved in the detection and repair of dsDNA breaks. The finding that a DNA ligsase IV syndrome patient, lacking the important ligase required for NHEJ repair, developed EBV-positive B-cell lymphoma is also intriguing if one assumes that the virus thrives in a situation where the balance between NHEJ and HR is dramatically shifted exclusively toward homologous repair [233]. In addition to containing HR proteins, EBV replication compartments have been shown to contain mismatch repair (MMR) proteins, including PCNA, RC-F, MSH2, MSH6, MLH1 and hPSM2, which are loaded on to the viral genome and copurified with the BMLR1 protein [234] and the RecQL helicase, which are associated with Zta [235]. The BGLF4 viral kinase interacts with the XPC protein, a member of a third DNA repair pathway, nucleotide excision repair. BGLF4 or XPC knockdown results in decreased viral replication [236,237], and BGLF4 expression enhances cellular XPC-mediated DNA repair in vivo [236].

The BGLF4 homolog in γ-MHV-68, encoded by ORF36, also plays a role in the induction of the DNA damage response by directly phosphorylating γ-H2AX, the dsDNA break sensor. This activation of γ-H2AX is further enhanced by ATM, and all three proteins are required for efficient replication of the virus in primary mouse macrophage cells [238]. However, the requirement for induction of a DNA damage response may be cell-type specific as fibroblast infection results in the inhibition of NHEJ/HR. This inhibition is mediated by the unique viral M2 protein, which, although sufficient to induce expression of ATM, also binds to ATM causing inhibition of the downstream effectors γ-H2AX, Nbs1 and 53BP. M2 is also capable of inhibiting nucleotide excision repair through interaction with the DDB1–COP9–cullin repair complex [239]. KSHV also encodes four interferon response factor-like proteins capable of blocking DNA repair pathways by the inhibition and degradation of p53 [240,241]. Consequently, neither p53 nor active ATM accumulates in KSHV-infected fibroblasts. By contrast, in the KSHV-infected lymphocyte cell line BCBL1, several DNA repair proteins were found bound to OriLyt and localized to replication compartments. These include RecQL helicase [242], the MMR proteins MSH2 and MSH6, NHEJ proteins DNA–PK_CS, Ku86 and Ku70, and poly-ADP ribose polymerase 1 (PARP1). PARP1 inhibitors were shown to diminish replication whereas hydroxyurea, which raises PARP1 activity, caused an increase in the DNA replication [242]. These observations provide compelling arguments that DNA recombination and repair activities play a critical role in the early stages of herpesvirus lytic replication.

Transcriptional requirements

There is mounting evidence in support of a role for RNA and transcription in the initiation of lytic herpesvirus replication. First, the lytic origin-binding proteins of both EBV and KSHV (Zta and Rta) are both transcription activators. It has also been known for quite some time that RNA polymerase II, TATA binding protein and TATA binding protein-associated factors [243] are recruited to HSV1 replication compartments [244,245]. What has become clearer recently is that transcription plays a major role in the selection of origins in mammalian cells [246]. All herepsvirus lytic origins consist of promoters containing transcription-factor binding sites important for replication. For example, varicella zoster virus OriS-dependent DNA replication and origin promoter transcription both require binding of the cellular factors Sp1 and Sp3 [247]. EBV replication is dependent on transcription of the BHRF1 promoter and, possibly, the BHLF1 promoter [122,248], independent of the gene product or even promoter sequence [121]. In the same way, KSHV and CMV OriLyt-mediated replication are also dependent on activation of their bidirectional promoters [60,90]. Several RNA species transversing herpesvirus origins have also been identified. These include OriS-RNA2, which overlaps HSV1/2 OriS [249,250], and a family of small bottom strand RNAs, which overlap the EBV OriLyt UEE [251]. The best-characterized lytic origin RNA, identified in CMV OriLyt, forms a persistent RNA–DNA hybrid structure [142] containing an RNA stem-loop sequence that is bound by the UL84 origin-binding protein [59]. Interestingly, an RNA–DNA hybrid has also been found at the human mitochondrial heavy-strand origin [252]. RNA may also play a regulatory role in latent replication of EBV at OriP [253,254]. These observations suggest a role for transcription and possible RNA itself in the initiation of lytic replication.

Additional factors involved in EBV lytic replication

Several additional viral proteins have also been shown to play important, although not always necessary, roles in enhancing lytic replication. These include the BGLF5 alkaline exo-nuclease [215] and BGLF4 kinase [237] described earlier. In addition to its role in promoting DNA recombination, the BGLF4 protein also localizes to replication compartments where it phosphorylates Zta [255], EA-D [256] and the MCM4–MCM6–MCM7 complex, which it also activates [257]. BGLF4 also phosphorylates EBV nuclear antigen 1, which remains expressed and bound to OriP during lytic replication [258], resulting in disruption of latent genome maintenance [259].

Uracil DNA glycosylases (UDGs) are also important since the inhibition of both viral and cellular UDGs significantly impairs lytic replication [260,261]. Studies involving the CMV homolog of EBV BKRF3 (UL114) suggest that UDGs are an integral part of the early-to-late replication switch mechanism of the virus, whereby incorporation of uracils into the viral genome followed by UDG and exonuclease activity may lead to strand breaks creating substrates for recombination-dependent replication [262]. This is an intriguing possibility given that uracil incorporation into HSV1 OriS would abrogate UL9 binding [263], perhaps mediating a switch from UL9-dependent to UL9-independent replication. Further proof of principle has been provided by demonstration of the ability of HSV1 UDG (UL2), processivity factor (UL42) and DNA polymerase (UL30) to cooperate with human AP endonuclease to create a DNA single-strand break in vitro [264]. Additional lines of evidence suggest that viral UDGs are not merely occasional repair proteins. These include the observation that:

• ▪Recombinant CMV genome lacking the viral UDG gene (UL114) did not accumulate more uracil compared with the wild-type virus [265];
• ▪The catalytic activity of viral UDG was very inefficient as compared with human UDG;
• ▪The viral processivity factor is required for the viral UDG to be loaded onto DNA [266].

Conclusion & future perspective

The most successful therapeutic interventions currently used against herpesvirus infection and associated diseases target the lytic replication of viral DNA. A better understanding of the enzymes and mechanisms involved in this process will likely yield additional drug targets and improved treatment options in the future. Our survey of the literature suggests that a large diversity of mechanisms are employed to generate the initiating nucleoprotein structures and host cell environment conducive to lytic replication. Nevertheless, these various mechanisms converge on common pathways that include highly conserved viral-replication enzymes and cellular factors involved in host cell DNA recombination, repair and replication. Other common requirements include the host cell-cycle arrest, a nuclear reorganization into replication compartments and a near-universal requirement for RNA transcription. We suggest that common elements, such as DNA hairpin structures bound to origin-binding proteins, transcription initiation factors and RNA transcripts, contribute to the formation of a higher-order structure that is recognized by the core viral DNA replication machinery (Figure 4). A major focus of future research will be investigating the mechanistic contributions of DNA repair proteins (e.g., viral and cellular endonucleases and recombinases) to the initiation and progression of lytic DNA replication. Certainly, the role of virus proteins on cell-cycle control have been investigated extensively, but the precise mechanism of cell-cycle rerouting used by herpesviruses deserves further attention. For γ-herpesviruses, the common use of C/EBP-α factors for lytic-cycle replication and cell-cycle arrest raises some important questions, including whether other viruses use this family of proteins to control their origins of replication, and whether this is coordinated with cell metabolic state and differentiation status, where C/EBP-α is known to play an important role. Finally, it will be necessary to determine whether RNA transcription contributes to lytic replication initiation. Active transcription may facilitate formation of an active prereplication complex by stimulating strand unwinding and torsional strain. Alternatively, transcription factors and RNA polymerase accessory factors may facilitate chromatin remodeling and replisome assembly. It will also be interesting to determine whether the transcribed RNAs (both coding and noncoding) contribute directly to origin protein assembly or function during lytic replication. These and other areas of future investigation suggest that exciting discoveries will soon emerge from further studies of herpesvirus DNA replication.

Figure 4. Hypothetical model of common structures formed by herpesvirus origins.

EBV, KSHV and HSV OriS are envisioned to form higher-order structures that include inverted repeat hairpins that bind viral (UL9, Zta) or cellular (C/EBP-α) proteins. Transcription complexes formed at both ends of OriLyts are envisioned to induce topological changes that promote DNA unwinding and templates suitable for viral DNA replisomes to assemble.

EBV: Epstein–Barr virus; HSV: Herpes simplex virus; KSHV: Kaposi’s sarcoma-associated herpesvirus; OriS: Repeated short-region origin of replication.

Executive summary.

Introduction

• ▪All herpesviruses share a common requirement of lytic replication to produce progeny virus. Although core replication enzymes are conserved among the family members, significant variation is observed at the level of initiation and origin-binding factors.
• ▪Lytic replication is a major source of all herpesvirus pathogenesis and contributes to cancer cell evolution during chronic infections of Epstein–Barr virus (EBV) and Kaposi’s sarcoma-associated herpesvirus.

Herpesviridae core replication proteins

• ▪All herpesviruses share a core set of highly conserved lytic-replication proteins, which can be substituted for one another in many cases.
• ▪In the case of EBV, these include a viral polymerase (BALF5, currently the major target of herpesvirus antivirals), polymerase processivity factor (BMRF1/EA-D), helicase–primase complex (BBLF4, BSLF1 and BBLF1/2) and a ssDNA-binding protein (BALF2).

Diverse group of origin-binding proteins

• ▪In contrast to the homologous core replication proteins, herpesviruses each encode very different lytic origin-binding proteins.
• ▪The EBV lytic origin-binding proteins are Rta and Zta, which also function as immediate-early transcriptional activators. Although not very similar to the other origin-binding proteins, it appears to share some intriguing overlapping functions.

Herpesvirus origins of lytic replication

• ▪Genetic experiments and plasmid replication assays have been used to identify important cis-acting viral elements required for lytic DNA replication.
• ▪Although EBV OriLyt and the lytic origins of other herpesviruses vary greatly in sequence composition, they share many tantalizing similarities including the presence of bidirectional promoters, CCAAT boxes and inverted repeat sequences bound by replication initiator proteins.

θ, rolling circle & recombination?

• ▪It has been proposed that herpesviruses copy their genomes via a ‘rolling-circle’ method of lytic replication. Several lines of evidence support this model. However, a simple rolling-circle mechanism does not adequately explain the highly branched DNA intermediates observed during replication.
• ▪This has led to a dual-mechanism model, where lytic replication is initiated via a plasmid or ’θ’ mode where copy number is enriched followed by a switch to rolling circle.

Role for cellular recombination & DNA damage-response proteins

• ▪It is likely that herpesvirus lytic replication involves a recombination mechanism. In addition to encoding their own proteins capable of promoting recombination, herpesviruses that interface with host cell proteins are involved in cellular DNA damage repair–recombination pathways.

Transcriptional requirements

• ▪There is mounting evidence in support of a role for RNA and transcription in the initiation of lytic herpesvirus replication. Herepsvirus lytic origins consist of promoters containing transcription-factor binding sites important for replication and several RNA species transversing hepesvirus origins, which may play a role in replication.

Additional factors involved in EBV lytic replication

• ▪Several additional proteins have also been shown to play important, although not always necessary, roles in enhancing lytic replication. These include endonucleases and uracil DNA glycosylases, which may provide insight into the replication mechanisms.

Conclusion & future perspective

• ▪The most successful therapeutic interventions currently used against herpesvirus infection and associated diseases target the lytic replication of viral DNA. A better understanding of the enzymes and mechanisms involved in this process will likely yield additional drug targets and improved treatment options in the future.
• ▪Three major focuses of future research will be investigation into the mechanistic contributions of DNA repair proteins (e.g., viral and cellular endonucleases and recombinases) to the initiation of DNA replication upon lytic reactivation, the use of the C/EBP family of proteins as origin controls and the role of RNA and RNA transcription in origin function.