Background
Epstein-Barr virus (EBV) is replicated once per cell-cycle as an episome in proliferating latently infected cells [
1,
2]. Episomal replication requires a viral sequence
in cis, termed
oriP, and a single viral protein EBNA1 [
3,
4].
OriP contains two sets of binding sites for EBNA1, the region of dyad symmetry (DS), that contains four sites of low affinity for EBNA1, and the family of repeats (FR) that contains twenty high-affinity sites for EBNA1 [
5,
6]. DNA synthesis initiates at DS, in a manner dependent upon the association of the cellular origin recognition complex (ORC) proteins and minichromosome maintenance (MCM) proteins with DS [
7‐
9]. Recent evidence indicates that EBNA1 recruits the ORC proteins to DS through an RNA-mediated interaction with ORC1 [
10].
FR functions as a plasmid maintenance and partitioning element [
11,
12]. FR from the prototypic B95-8 strain of EBV contains 20 high-affinity sites for EBNA1, which binds each of these sites as a dimer [
13,
14]. EBNA1 bound to FR tethers viral episomes or
oriP plasmids to cellular chromosomes [
15‐
19]; an association that facilitates the plasmids to piggy-back into daughter cells at each metaphase [
20,
21]. In addition to its role in genome partitioning, two-dimensional gel analysis by Schildkraut and co-workers has indicated that the migration of replication forks through FR is attenuated, so that for the circular EBV genome or an
oriP-plasmid, the bidirectional replication fork that initiates at DS is terminated at FR [
22]. This ability of EBNA1 bound to FR to attenuate replication forks has been recapitulated in biochemical assays performed
in vitro; such assays reveal that DNA binding domain of EBNA1 bound to FR impede the migration of replication forks from an SV40 origin on the same template [
23].
Using assays for transcription activation and plasmid maintenance, we have examined the binding site requirements for EBNA1 in the EBV FR in detail [
24]. Our analyses indicated that although the wild-type FR contains 20 binding sites, plasmids with 10 binding sites are maintained far more efficiently in colony formation assays than the former (
ibid). A similar finding has been reported for deletion mutants constructed within the natural FR, in that a plasmid with nine binding sites replicated more efficiently than a plasmid with twenty binding sites [
25]. Thus these results concur in that the wild-type number of EBNA1-binding sites in FR limits the replication of
oriP-plasmids by acting
in cis.
In this study, we have examined the mechanism by which the wild-type number of binding sites limits the replication of oriP-plasmids. Our results indicate that EBNA1 bound to FR limits replication by impeding the migration of replication forks from DS. In addition, we have determined that EBNA1 bound to FR severely impairs the migration of transcription forks through FR. We discuss both these findings in the context of the stable replication of EBV episomes.
Methods
Bacterial strains and plasmid purification
All plasmids were propagated in the
E. coli strains DH5α, MC1061/P3, or STBL2 (Invitrogen, Carlsbad, CA). Plasmids used for transfection were purified on isopycnic CsCl gradients [
26].
Plasmids
Plasmids AGP73, and AGP74 have been described previously [
24], and contain 10 and 20 EBNA1-binding sites in the FR respectively. These plasmids are constructed in the backbone of pPUR, and also contain EBV's DS and the EBV sequences between FR and DS. AGP81 contains 40 EBNA1-binding sites in FR and was constructed by dimerizing the FR in AGP74. AGP82 contains 80 EBNA1-binding sites in FR and was constructed by dimerizing the FR in AGP81. AGP83 has been described previously and is a control plasmid that only contains DS and completely lacks FR. AGP212, and AGP213 contain 20 EBNA1-binding sites split into two FRs each containing ten binding sites as described in the Results section. They were constructed as derivatives of AGP73. AGP212 was constructed by recovering an
Mfe I-
Eco RV fragment containing FR from AGP73 and inserting it into the
Eco RI-
Bam HI sites of that plasmid. AGP213 was constructed by inserting an
Eco RV-
Acc 65I fragment from AGP73 into the
Acc65 I site of the same plasmid. Plasmid 2380 contains wild-type
oriP cloned in pPUR, and was a gift from Bill Sugden. Plasmids AGP39, AGP40, and AGP41 were constructed as derivatives of pRSVL, by inserting 10, 20, or 40 EBNA1 binding sites between the end of the luciferase open reading frame and the SV40 polyadenylation signal in that plasmid. Plasmid 1606 has been described previously and expresses the large T antigen of SV40 under the control of the CMV immediate early promoter [
27]. Plasmid 1160 has been described previously and expresses the DNA binding domain of EBNA1 under the control of the CMV immediate early promoter [
28]. The empty expression vector, pcDNA3, was used as a control plasmid. Plasmid 2145 has been described previously and expresses EGFP under the control of the CMV immediate early promoter [
17].
Cell culture and transfections
The human cell line 293 [
29], and its EBNA1-expressing derivative, 293/EBNA1, were used in this study. Both cell-types were grown in DMEM supplemented with 10% fetal bovine serum. G418 was added at a concentration of 200 mg/L to the media for 293/EBNA1 cells. Cells were grown at 37°C in a humidified 5% CO
2 atmosphere. Plasmids were introduced into cells by the calcium phosphate method as described previously [
17,
18,
24]. Transfections were normalized by the inclusion of a CMV-EGFP expression plasmid, 2145, in each transfection. Upon harvest, a fraction of the cells were profiled using a Becton-Dickinson FACSCalibur. Transfection efficiency was measured as the fraction of GFP-expressing, live cells quantified using CellQuest software from Becton-Dickinson (Franklin Lakes, NJ).
Ten μg of AGP74 or an equivalent number of moles of plasmids AGP73, AGP81, 2380, AGP82, AGP83, AGP212, and AGP213, were co-transfected with 1 μg of 2145 into 1 × 10
7 293/EBNA1 cells on a 10 cm dish. Cells were split eight hours post-transfection so that they would not be confluent at 48 hours post-transfection, at which time cells were harvested, FACS profiled to measure GFP expression, and re-plated in duplicate at 2 × 10
5, 2 × 10
4, and 2 × 10
3 GFP-positive, live cells per culture dish. Cells were placed under selection with 0.5 μg/ml puromycin four days post-transfection. After two weeks of selection, the resulting puromycin-resistant colonies were fixed with formamide and subsequently stained with methylene blue. Colonies that were at least 2 mm in size were scored as positive. Colonies were counted using a colony counting macro written for NIH Image as described previously [
17,
18].
Southern hybridization analysis to assess plasmid replication
Ten μg of AGP74 or an equivalent number of moles of plasmids AGP73, 2380, AGP212, and AGP213 were co-transfected with 1 μg of 2145 into 1 × 10
7 293/EBNA1 cells on a 10 cm dish. Cells were placed under puromycin selection 48 hours post-transfection. After three weeks of selection, episomal DNAs were extracted from cells in puromycin resistant colonies that were pooled. Episomal DNAs were extracted from 2 × 10
7 – 10
8 puromycin-resistant cells as described previously [
11,
30]. Extracted DNAs were digested with 200 units of
Dpn I, 20 units of
Bam HI, and 20 units of
Xba I in a final volume of 100 μl overnight at 37°C. Restriction endonucleases were purchased from New England Biolabs (Beverly, MA), and used as per the manufacturer's instructions. Digestions were extracted with phenol:chloroform (1:1), precipitated and electrophoresed on a 0.8% agarose gel. DNAs were transferred from the gel to Hybond membrane (Amersham, Buckinghamshire, UK) using an Appligene vacuum transfer apparatus (Boekel Scientific, Feasterville, PA). Radioactive probes were prepared by the incorporation of α-
32P-dCTP (6000 Ci/mmol) (Amersham) during Klenow synthesis using random primers and
Pst I-digested AGP83 as template. Probe specific activities ranged from 1 × 10
9 cpm/μg to 3 × 10
9 cpm/μg. Southern hybridization was performed as described by Hubert and Laimins [
31,
32]. Southern blots were visualized and quantified by phosphorimage analysis using a Molecular Dynamics Storm phosphorimager (Molecular Dynamics, Sunnyvale, CA).
Transfection of linear plasmid DNAs to assess replication fork migration in vivo
Ten μg of Pvu II-linearized AGP73 or AGP74 was transfected into 293/EBNA1 cells as described above along with 1 μg of 1606. Hirt extracts were prepared from 2 × 107 transfected cells 14 – 16 hours post-transfection and digested exhaustively with Dpn I (200 units). The digested extracts were then digested with Hin dIII (10 units) &Acc 65I (10 units) to release a 1063 bp fragment between the SV40 origin and FR, and with Bsr GI (10 units) &Spe I (20 units) to release a 637 bp fragment that lies immediately after FR. The digested products were separated on a 1.5% agarose gel electrophoresed in 0.5× TBE, and transferred to Hybond membrane and probed as described above. Probe was synthesized using random primers and the Hin dIII-Acc 65I fragment, as well as the Bsr GI-Spe I fragment as template. In control experiments, probes were hybridized against purified fragments to confirm that the Bsr GI-Spe I fragment bound approximately two-thirds as much probe as the Hin dIII-Acc 65I fragment.
Transcription reporter assays
100 ng of pRSVL [
33], or an equivalent number of moles of AGP39, AGP40, or AGP41 was co-transfected with 1 μg of 2145 and 10 μg of pcDNA3 or 10 μg of 1160 into 293 cells. Cells were split eight hours post-transfection so that they would not have reached confluence when harvested 72 hours post-transfection. A fraction of the harvested cells were then counted twice using a Coulter counter, and FACS profiled to normalize for the fraction of live transfected cells. The remainder of the cells were pelleted, and lysed in reporter lysis buffer (provided along with a luciferase assay kit from Promega, Madison, WI) at a concentration of 1 × 10
5 cells/μl. Lysates were spun for 5 minutes at 1000 g to remove nuclei, and then frozen at -80°C until assay. Luminescence assays were performed as per manufacturer's instructions, using a Zylux FB 15 luminometer.
RT-PCR analysis to measure migration of transcription forks through FR
Total RNA was extracted from transfected 293 cells using the SV Total RNA Isolation System from Promega (Madison, WI). PolyA+ RNA was extracted from transfected 293 cells using the PolyATract mRNA Isolation System from Promega (Madison, WI). Either 5 μg of total RNA or 1 μg of polyA RNA was used in RT-PCR reactions using the following primers to detect firefly luciferase:
AGO83: 5' GGAATACTTCGAAATGTCCG
AGO84: 5' TCATTAAAACCGGGAGGTAG
Control RT-PCR reactions amplifying the glyceraldehyde phosphate dehydrogenase (GAPDH) transcript were performed using the following two primers:
AGO81: 5' CTCAGACACCATGGGGAAGGTGA
AGO82: 5' ACTTGATTTTGGAGGGATCTCG
RT-PCR reactions were performed using the AccessQuick one-tube RT-PCR System purchased from Promega (Madison, WI).
Discussion
In this study we have demonstrated that the wild-type number of EBNA1-binding sites in EBV's FR region is sub-optimal for the efficient replication of
oriP-plasmids. A plasmid with ten binding sites in FR formed colonies more efficiently than a plasmid with the wild-type number of 20 binding sites. Increasing the number of binding sites in FR beyond 20 further decreased the efficiency of replication (Table
1). These results corroborate those of Leight and Sugden who have demonstrated that an
oriP-plasmid with a deletion that removes approximately one-half of FR is replicated more efficiently than a plasmid with wild-type FR [
25].
We have tested two models to explain why plasmids with fewer binding sites in FR are replicated more efficiently than plasmids with the wild-type number of binding sites. Our results support a model wherein EBNA1 bound to FR impedes the progression of replication forks that originate from DS. It was determined that this effect correlates with the number of contiguous EBNA1 binding sites in FR. Attenuation of fork migration is not readily detected with ten contiguous sites, but is easily observed with 20 contiguous binding sites. As has been observed previously
in vitro [
23], we found that EBNA1 bound to FR also impedes replication forks from the SV40 origin within transfected cells. The SV40 origin was used for this analysis because it fires multiple times in a single cell-cycle, permitting facile evaluation of the reduction in fork migration. The major difference between the SV40 replication fork and replication forks that initiate from DS lies in the nature of the leading strand helicase. The hexameric large T-antigen helicase in the SV40 replication fork has approximately the same mass as the hexameric MCM helicase present at replication forks that initiate from DS [
7]. Both forks progress at similar rates, with elongation being estimated at approximately 100 bp/min for the SV40 replication fork [
42,
43], and at between 10 – 50 bp/min for EBV replication [
44]. Given the similar biophysical characteristics of both forks, we believe that EBNA1 bound to FR will impede the progression of replication forks that fire from DS in a manner dependent on the number of binding sites.
Our data indicates that split-FR plasmids containing two FRs with ten binding sites each are replicated more efficiently than plasmids containing a single FR with twenty contiguous EBNA1 binding sites (Table
2). EBNA1 bound to FR tethers
oriP-plasmids to chromosomes to facilitate their maintenance and partitioning in proliferating cells [
17‐
19]. The efficiency of this process is dependent upon the number of binding sites in FR, such that an EBNA1 mutant which is partially defective in chromosomal association can be rescued by increasing the number of binding sites in FR [
18]. However, with 20 contiguous sites, this increase in partitioning efficiency is offset by a decrease in replication efficiency. Splitting the 20-binding site FR into separated FRs with ten binding sites each, no longer impedes replication, but retains the advantage of having 20 binding sites for efficient
oriP-plasmid partitioning. The data obtained with AGP212 and AGP213 (Table
2) also indicates that the replication factor titration model proposed previously is unlikely. Both these plasmids contain 20 EBNA1 binding sites and replicate more efficiently than a plasmid that contains ten binding sites. Were the titration model to be correct, replication of these plasmids would be less efficient than replication of a plasmid with ten EBNA1 binding sites in FR.
The ability of EBNA1 to impede replication fork migration likely impacts replication of EBV genomes. Besides DS, there are other replication origins on the EBV genome also used during latency [
45], such as an origin that lies in the
Bam HI-A fragment [
46,
47]. It is known that collision of replication forks can lead to fork collapse, and the consequential generation of double-stranded breaks (DSBs) [
48]. Such events can lead to irregular recombination events, and a large number of DSBs causes apoptosis [
49‐
51]. We propose that EBNA1 bound to FR acts as a buffer to prevent two replication forks from running into each other and thereby protects cells latently infected by EBV from undergoing apoptosis as a consequence of DSB generation.
There is a striking parallel between the function of EBNA1 at FR and TTFI at
Sal repeats that terminate ribosomal DNA replication. Both proteins impede the progression of replication forks dependent on the number of binding sites for the protein on the template DNA [
52]. Additionally, just as TTFI bound to the
Sal repeats blocks the progression of transcription forks and terminates them [
53], we have found that EBNA1 bound to FR blocks the progression of transcription forks in a manner dependent upon the number of binding sites (Figure
5). In its natural chromosomal context the ribosomal DNA replication fork block is required for the proper termination of rRNA transcripts. Within the EBV genome, the EBER RNA genes are immediately 5' of FR and transcribed toward it [
34]. The EBERs are pol III transcripts [
54,
55], and it is now known that some cellular pol III transcripts are terminated by pol II transactivators acting as transcription fork blocks [
56]. On this basis, we speculate that EBNA1 bound to FR participates in the proper termination of EBER RNAs. It is also possible that FR prevents transcription forks emanating from the EBER genes colliding with replication forks emanating from DS. Similar to collisions between replication forks, such collisions also cause replication-fork collapse, with the consequent pro-apoptotic generation of DSBs.
In conclusion, there are several reasons for EBV to have an FR that is sub-optimal for plasmid replication. It is clear that EBNA1 bound to FR activates transcription from multiple viral promoters [
57‐
59], a property of EBNA1 necessary for naïve B-cells to be immortalized by EBV [
60]. We and others have demonstrated that ability of EBNA1 to activate transcription is proportional to the number of binding sites in FR [
24,
61]; EBNA1 bound to 20 binding sites activates transcription approximately two to three times as well as EBNA1 bound to ten binding sites [
24]. Thus, while the number of EBNA1-binding sites in FR is sub-optimal for replication of
oriP-plasmids, this number of binding sites is likely necessary for EBNA1 to transactivate effectively. It is also intriguing that when bound to 20 binding sites, EBNA1 functions effectively as a transcription and replication fork-block, leading us to conjecture that the latter activity protects latently infected cells by preventing DNA damage resulting from collisions between a replication fork originating at DS, and transcription or replication fork-blocks emanating from elsewhere in the EBV genome.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AA was responsible for experimental design, conducting experiments, and writing the manuscript. SA was responsible for experimental design, and conducting experiments. AW was responsible for conducting experiments. GS was responsible for conducting experiments. RBL was responsible for experimental design, and writing the manuscript.